Quark matter influence on observational - Instituut

Transcription

Quark matter influence on observational - Instituut
Quark matter influence on observational
properties of compact stars
Sjoerd Hardeman
Master’s thesis Theoretical Physics & Astronomy and Astrophysics
Supervisors:
Dr. Daniel Boer
Dr. Rudy Wijnands
Three quarks for Muster Mark!
Sure he hasn’t got much of a bark
And sure any he has it’s all beside the mark.
But O, Wreneagle Almighty, wouldn’t un be a sky of a lark
To see that old buzzard whooping about for uns shirt in the dark
And he hunting round for uns speckled trousers around by Palmerstown Park?
Hohohoho, moulty Mark!
You’re the rummest old rooster ever flopped out of a Noah’s ark
And you think you’re cock of the wark.
Fowls, up! Tristy’s the spry young spark
That’ll tread her and wed her and bed her and red her
Without ever winking the tail of a feather
And that’s how that chap’s going to make his money and mark!
From: Finnegans Wake
James Joyce
Astronomical Institute ‘Anton Pannekoek’
Faculty of Science
Universiteit van Amsterdam
Institute for Theoretical Physics
Department of Physics and Astronomy
Faculty of Sciences
Vrije Universiteit
Abstract
Observations of compact stars show again and again that the theory describing these objects is far
from complete. In this thesis the possibility that free quark matter is present in such objects will be
considered, studying the observational consequences of quark matter in compact stars. Free quark
matter can be present at high densities as the coupling constant of the strong force becomes weaker
at high energies. At low energies quarks are confined into hadrons, but at high energies the coupling constant might become so weak that deconfinement occurs. Compact stars thereby form a very
interesting laboratory to study matter that cannot be made on earth.
One of the main differences between quark matter and hadronic matter is its different relation of
pressure and energy, expressed in the equations of state. Compact stars made of quark matter are
smaller, and the mass-radius relation shows an increasing mass with radius for most of the domain
of allowed masses. This in sharp contrast with hadronic matter where the mass-radius relation has a
negative slope for the entire mass domain. Furthermore, magnetic fields behave differently in quark
matter and hadronic matter. Where hadronic matter is a superconductor, quark matter behaves only
as a very good conductor. This has important influences on changes in the electromagnetic field, for
example when the compact star is precessing. We will see that study of the behaviour of the magnetic
field provides us with a very good probe of the interior of a compact star.
i
Preface
This thesis concludes of one and a half year of research on observational properties of quark stars,
based on quantum field theoretical calculations of quark matter. I have done my research at the
‘Anton Pannekoek Institute for Astronomy’ of the University of Amsterdam and at the department
of Theoretical Physics of the Vrije Universiteit, as my Master’s research the combined Masters for
Theoretical Physics and Astronomy and Astrophysics.
The theory of colour interactions (QCD) has the peculiar behaviour that the interaction strength
declines for higher energies. Therefore, calculations at high energies are much less challenging than
calculations at low energies. At the end of the seventies Baluni (1978), Freedman and McLerran
(1977c) already calculated the QCD thermodynamic potential at high densities. Their results lead to
the suggestion by Witten (1984) that in the early universe densities may have been such that quark
stars have emerged. Soon after it was realised that the very high temperatures in the early universe
would have quickly destroyed these stars (Alcock and Farhi, 1985).
The idea that at high densities quark behave as weakly interacting particles, however, has since
been considered a real possibility for the dense objects remaining after the death of a massive star.
These objects are generally known as neutron stars, although they might contain quark matter. I will
use the term compact star for these objects, in contradiction with the usual definition of compact star
which also includes white dwarf stars and black holes. Usually compact objects as defined here are
referred to as neutron stars, a very unfortunate name that can lead to ambiguities. I will reserve the
term ‘neutron star’ explicitly for stars made of only baryonic matter. The term ‘quark star’ will be
used only for compact stars made of deconfined quark matter. For objects known to contain both I
will use the term ‘hybrid star’.
I will first discuss the observational properties of these objects and discuss the formation and
evolution in chapter 1. Followed by a discussion of the equations governing compact star structure
in chapter 2. In chapter 3 I will discuss some concepts quantum field theory and more specifically
quantum chromodynamics. Using these concepts, I will show some details of the derivation of the
quark matter thermodynamic potential in chapter 4. Based on this thermodynamic potential, I have
performed calculations of stellar models. My results can be found in 5.
One particular property of quark matter is colour superconductivity. Although of little influence on
the mass-radius relations, it has profound influence on the behaviour of magnetic fields. The theory of
both superconductivity and colour superconductivity is shortly reviewed in chapter 6. Based on this,
I will discuss magnetic field behaviour and its observational consequences for both neutron stars and
quark stars in chapter 7. All conclusions are summarised in chapter 8.
Sjoerd Hardeman
Amsterdam, March 30, 2007
iii
Contents
Abstract
i
Preface
iii
1
2
3
Introduction
1.1 Compact star observations . . . . . . . . . . . . . . . . . .
1.1.1 Pulsars . . . . . . . . . . . . . . . . . . . . . . . .
1.1.2 X-ray binaries . . . . . . . . . . . . . . . . . . . . .
1.1.3 Surface radiation of compact stars . . . . . . . . . .
1.1.4 Free precession . . . . . . . . . . . . . . . . . . . .
1.2 Formation and evolution of compact stars . . . . . . . . . .
1.2.1 Stellar evolution and the formation of compact stars
1.2.2 Supernovae . . . . . . . . . . . . . . . . . . . . . .
1.2.3 Magnetic fields . . . . . . . . . . . . . . . . . . . .
1.2.4 Cooling of compact stars . . . . . . . . . . . . . . .
Structure of compact stars
2.1 General structure equations . . . . . . . . . . . . .
2.1.1 Polytropes . . . . . . . . . . . . . . . . .
2.2 Modelling stars . . . . . . . . . . . . . . . . . . .
2.2.1 Models of neutron stars and quark stars . .
2.2.2 Solving the models . . . . . . . . . . . . .
2.2.3 Equations of state . . . . . . . . . . . . . .
2.2.4 Effect of rotation on compact star structure
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Quantum chromodynamics at high densities
3.1 Symmetry groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Quantum field theory . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.1 Instantons . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3 Confinement and asymptotic freedom . . . . . . . . . . . . . . . . . .
3.3.1 Quantum chromodynamics at low energies . . . . . . . . . . .
3.3.2 Phase transitions in dense matter . . . . . . . . . . . . . . . . .
3.4 The running coupling constant in QCD . . . . . . . . . . . . . . . . . .
3.4.1 The running of the coupling constant . . . . . . . . . . . . . .
3.4.2 The running of the quantum chromodynamics coupling constant
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
1
4
6
6
7
7
8
10
12
.
.
.
.
.
.
.
15
15
16
18
18
19
20
21
.
.
.
.
.
.
.
.
.
23
23
24
26
27
28
29
30
31
31
v
CONTENTS
4
.
.
.
.
.
33
33
35
35
37
39
5
Mass-radius relation of quark stars
5.1 Numerical stellar models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.1 The MIT bag model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 Numerically solving models of quark stars . . . . . . . . . . . . . . . . . . . . . . .
5.2.1 The method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.2 Equations of state used in the model . . . . . . . . . . . . . . . . . . . . . .
5.2.3 The programme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3 The equation of state from perturbative quantum chromodynamics . . . . . . . . . .
5.3.1 Equation of state of three massless quarks . . . . . . . . . . . . . . . . . . .
5.3.2 Equation of state for one massive quark flavour . . . . . . . . . . . . . . . .
5.4 Numerical solutions to quark star models using a perturbative QCD equation of state
5.4.1 Quark star models using an equation of state of three massless quarks . . . .
5.4.2 Quark star models using an equation of state of one massive quark flavour . .
5.5 Conclusions on the mass-radius relation . . . . . . . . . . . . . . . . . . . . . . . .
43
43
44
46
46
47
48
48
48
50
53
53
55
56
6
Theory of superconductivity
6.1 Superconductivity in neutron matter . . . . . . . . . . . . . . . . . . . . . . . . . .
6.1.1 Type I superconductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.1.2 Type II superconductivity . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2 Colour superconductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2.1 Colour-Flavour Locked phase . . . . . . . . . . . . . . . . . . . . . . . . .
6.2.2 Two-quark superconducting phase . . . . . . . . . . . . . . . . . . . . . . .
6.2.3 One flavour pairing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2.4 Other colour superconducting phases and the quantum chromodynamics phase
diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
59
59
60
61
63
64
64
65
65
Magnetic fields in hybrid stars and quark stars
7.1 Neutron star magnetic behaviour . . . . . . . . . . . . . . .
7.1.1 Fluxoid pinning . . . . . . . . . . . . . . . . . . . .
7.1.2 Quark matter influence on magnetic fields . . . . . .
7.2 Consequences of quark matter for pulsar timing observations
7.2.1 Glitches . . . . . . . . . . . . . . . . . . . . . . . .
7.2.2 Precession . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
67
67
67
70
72
72
72
Conclusions and discussion
8.1 Mass-radius relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.2 Quark matter and superconductivity . . . . . . . . . . . . . . . . . . . . . . . . . .
75
75
77
7
8
Thermodynamics of degenerate quark matter
4.1 Fermi-Dirac distribution . . . . . . . . . . . . .
4.2 The MIT bag model . . . . . . . . . . . . . . . .
4.3 Finite temperature field theory . . . . . . . . . .
4.3.1 The thermodynamic potential . . . . . .
4.3.2 Thermodynamic quantities of a quark gas
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
A Derivation of the TOV equation
79
B Notation
83
vi
CONTENTS
Acknowledgements
85
Samenvatting
87
vii
CHAPTER 1
Introduction
In this chapter, some of the general properties of compact stars will be discussed. Throughout this
thesis, I will use the term compact star for an object that can either be a quark star or a neutron star. If
the term ‘neutron star’ or ‘quark star’ is used, I intent to discuss objects with a known matter content.
This is different from the usual definitions. Compact stars normally includes black holes and white
dwarfs, while the term ‘neutron star’ is usually used for all objects with radii ∼ 10 km, masses ∼ M
and are not black holes.
In the first part of this chapter the observational properties of compact stars are discussed. The
second part covers the theory of compact star formation and evolution to explain some of the observations.
1.1 Compact star observations
First we will consider some observational features of compact stars which can be used to distinguish
between different neutron star and quark star models. Compact stars are most often observed as
pulsars, spinning stars with strong magnetic fields. Due to the spinning field pulsars produce very
strong radiation, and can therefore easily be observed.
Even more interesting in this sense are the pulsar-white dwarf binaries and the double compact
stars, of which one of the stars is a pulsar. The radiation produced by pulsars is closely related to the
spin and therefore almost exactly periodic. When altered by general relativistic effects, these objects
allow for very accurate measurements.
Other mechanisms of producing electromagnetic radiation are found in binary systems of a normal
star with a compact star companion. If mass transfer occurs powerful electromagnetic radiation is
produced. Using that these objects reside in binary systems masses can be determined with reasonable
accuracy.
Finally, in Rosat surveys, isolated compact stars have been found (Agüeros et al., 2006) by their
X-ray emission. These objects produce no observable pulsar radiation.
1.1.1
Pulsars
The first pulsars were discovered by Jocelyn Bell-Burnell and Anthony Hewish (Hewish et al., 1968,
Pilkington et al., 1968). Pulsars are objects producing highly periodic pulses of electromagnetic radiation (hence the name: pulsating star). Very soon it was realised that these objects were associated with
rotating compact stars. Strong evidence for this conclusion came with the discovery of the Vela and
1
Chapter 1: Introduction
Crab pulsars (see figure 1.1), associated with recent supernovae (see section 1.2.2). In Glendenning
(2000, ch. 5) a historical overview is given.
When a star collapses, magnetic flux preservation predicts a strong increase in magnetic fields.
Also, angular momentum conservation predicts high spin rates. A spinning compact star with a strong
magnetic field will produce very strong low frequency electromagnetic radiation. Via a not fully
understood mechanism, this leads to the production of observable electromagnetic radiation from the
magnetic poles of the compact star.
In general, the magnetic poles and geographic poles do not coincide. As radiation is emitted in
beams from the magnetic poles, these beams will sweep around since these poles are spinning with
the compact star. If you happen to be in the path of such a beam, you will observe periodical emission
from this beam, with the spin frequency of the compact star as the period.
The emission of the electromagnetic waves
requires huge amounts of energy. The energy
radiated this way is drawn from the spin frequency, as ”dragging” the magnetic field through
the vacuum provides a resistance and slows the
star down. Using this model, magnetic fields can
be calculated from spindown rates, the rate of
change of the compact star spin. Assuming a
dipole field the spindown rate is given by (Bhattacharya and Srinivasan, 1995)
r
3I
B=
PṖ
(1.1)
2
8π R6 sin2 α
This formula can be used to draw lines of equal
field strength in the P − Ṗ diagram (figure 1.3).
For normal pulsars this suggests field strengths
∝ 108 T. These magnetic fields are of the order of the expected magnetic field in case of flux
preservation as will be discussed in section 1.2.3,
Figure 1.1: The Crab Nebula in X-ray. Image from
with the exception of magnetars and millisecond
the Chandra X-ray Satellite NASA/CXC/SAO (impulsars.
age from 1999)
The energy of the radiation has to come from
its angular momentum. It cannot come from
a decay of the electromagnetic field. Besides
the theoretical reasons provided hereafter, the
energetics of pulsar radiation match the energy
drawn from rotation very well (Glendenning,
2000, and references therein).
Neutron stars are thought to possess in their
core a superconducting proton fluid, and an alFigure 1.2: Spindown rate the Vela pulsar. Just
most perfect conducting electron fluid. Superafter a glitch the spindown rate is higher than norfluids do not allow magnetic fields to penetrate.
mal (image from Lyne and Graham-Smith, 2005)
This is known as the Meissner effect. Depending
on the type of superconductivity, the magnetic
field will form fluxoids or non-superconducting domains to penetrate the superconductor. In order to
lower the magnetic field of such a system, these fluxoids and domains have to move to the surface of
2
1.1 Compact star observations
the superconductor in order for the field lines to reconnect and decay. However, a moving magnetic
field will produce an electric field. The electron fluid then generates an opposed field preventing the
vortex or domain from moving. This mechanism ensures very stable magnetic fields. It is discussed
in much more detail in chapter 7.
One important observational feature of young pulsars are the glitches. As described above, pulsars
spin down due to the breaking effect of the pulsar mechanism. However, sometimes, for very short
periods, the star actually spins up. After this sudden change in spin the star relaxes exponentially to a
new spin-down rate, as shown in figure 1.2.
Millisecond pulsars
Millisecond pulsars are fast spinning pulsars
with a low magnetic field ∝ 104 T. These pulsars
are thought to be created from normal pulsars
in binary systems, spun up by matter accreted
from their companion stars (see section 1.1.2).
This model is supported by the fact that most of
the millisecond pulsars are found in binary systems with a white dwarf companion. Up to today, there are no binary pulsar systems in which
one of the two companions is a millisecond pulsar (Thorsett and Chakrabarty, 1999). This supports the idea that the millisecond pulsar is spun
up due to mass transfer. In systems with a more
massive companion, the faster evolution of the
companion will leave less time for spinning up
the compact star (Bhattacharya and Srinivasan,
1995). There are also a few millisecond pulsars
which do not have a companion. It is thought
that in these so-called black widow systems the
pulsar beam passes over the location of the original companion. Calculations show that it is possible that this beam evaporates the companion
star (Bhattacharya and Srinivasan, 1995). ObFigure 1.3: Diagram of period versus spindown
servational support is found by Wolszczan and
rate. Indicated are the field strength lines in Gauss
Frail (1992), who observed a planetary system
(104 G = 1 T). Points marked by a circle are biin PSR1257+12, made from the debris of evapnary systems, Stars represent magnetars (see secoration.
tion 1.1.1) The line marked by 1010 yr indicates the
During the accretion, a mechanism for reHubble time, the spin up line indicates how a normal pulsar should evolve to become a millisecond
ducing the magnetic field must have been at
pulsar (image from Manchester, 2006)
work. In Mandal et al. (2006) an explanation
for this decay is sought in the presence of quark
matter. They suggest the minimum observed fields ∝ 104 T can be explained assuming a quark star
in a superconducting phase. However, their suggestion is in contrast with the result of Alford et al.
(2000) that colour superconducting quark matter is not electromagnetically superconducting. Another
explanation can be provided by a locking of magnetic field lines to rotational vortices, as is expected
to happen in some superconductors. These results will be explored further in chapters 6 and 7. In
3
Chapter 1: Introduction
binary systems spin periods can be severely altered, with the longest period observed at almost 1000
seconds (Bhattacharya and Srinivasan, 1995). In such a spindown a large portion of the magnetic field
might be driven out. In contrast to this result observations show that in the slowly spinning pulsars in
binary systems with a massive companion, the high mass X-ray binaries, the fields are still ∼ 108 T,
as in ordinary pulsars (Coburn et al., 2006).
Magnetars
Magnetars are compact stars with a very strong magnetic field ∝ 1010 T. The stars are characterised
by their high spin down rates and X-ray emission. It is impossible to explain the emission from these
systems by rotation energy loss only. It is thought that the decay of the electromagnetic field also
contributes, and dominates the spindown emission (Kouveliotou et al., 1998). Magnetars are marked
by stars in the upper-right of the P − Ṗ diagram (figure 1.3).
Magnetic fields of magnetars are stronger than the critical field above which baryon matter becomes non-superconducting. Ohmic dissipation, the conversion of electric currents to heat through
Ohmic resistance, can therefore be a mechanism for field decay and energy generation. As discussed
in chapter 6, normally conducting nuclear matter has an extremely high electron conductivity. This
would lead to very long decay times for the magnetic fields, making Ohmic dissipation not a very
efficient mechanism for field decay.
Magnetars are often associated with quark matter. In Niebergal et al. (2006) quark superconductivity is used to explain the large magnetic fields. However, as discussed in section 6.2 and in
Alford et al. (2000) the main colour superconducting phases are not superconductors with respect to
electromagnetic fields. Therefore I consider this explanation unlikely.
Magnetars sometimes present giant X-ray outbursts. So far three have been observed (SGR
0526-66, SGR 1806-20 and SGR 1900+14). Israel et al. (2005) has discovered quasi periodic signals in these outbursts, that were attributed to star quakes (Strohmayer and Watts, 2005, Watts and
Strohmayer, 2006). This model only works for a crust thickness too large for a pure quark star. The
authors therefore ruled out that magnetars are quark stars. In contradiction to this, Levin (2006) used a
toy model to show that the model used suggests very strong damping of the high frequencies. As these
frequencies have been observed, the author considers the star quake model unlikely. His suggestion is
that the quasi periodic signals are associated with wave phenomena in the magnetar magnetic field.
1.1.2
X-ray binaries
Many compact stars have been found in binary systems in which mass transfer occurs. Due to the
very strong gravitational field of a compact star, up to 30 percent of the rest energy of the infalling
mass can be converted to energy. This leads to very high temperatures and causes these objects to be
mainly visible in X-rays. This allows for easy observation of compact stars in these binaries.
The binary has a very strong influence on the compact star. First, stars in a binary system will
evolve very different from a solitary star, as mass transfer can seriously alter the initial masses of
stars. This is demonstrated in the Algol system, in which the lighter star has evolved much further,
while it is commonly accepted that the lifetime of a star is inversely related to its initial mass. Based
on the mass-luminosity relationship L ∝ M03.5 (Close et al., 2005) with L the luminosity and M0 the
initial mass one can obtain T max ∝ M0−2.5 for the lifetime T max . For the last step the assumption is
made that the available fuel of a star is proportional to its mass by LT max ∝ M. The explanation for
this system is that initially the lighter star was more massive than its companion, but that accretion
has changed this situation dramatically.
4
1.1 Compact star observations
Accretion
Accretion occurs when mass from a normal star is transported via some mechanism to the compact
star. There are two different mechanisms: wind accretion and Roche Lobe overflow.
Roche lobe overflow occurs when the companion star grows larger than its Roche lobe,
the maximum equipotential surface in which the
gravity of the star dominates. This can be understood from a corotating potential
φ=−
GM1
GM2
1
−
− (Ω × r)2
r − r1 r − r2 2
(1.2)
with M1 and M2 the masses of the objects and Ω
the angular momentum. In figure 1.4 the potential is plotted for M1 /M2 = 2.
Conservation of angular momentum dictates
the inflow of matter via an accretion disk. In
such a disk, matter is transported inwards while
its angular momentum is transported outwards at
the expense of a small amount of matter. However, some of the angular momentum will reach
the compact star and increase the angular momentum of the object. This can go directly, by
matter falling on the compact star carrying angular momentum, or indirectly, by torques from
tidal effects.
Wind accretion happens when the companFigure 1.4: The potential of two rotating masses
ion star experiences a mass loss due to a stellar
with a mass ratio of 2:1
wind. This mass is blown out of the gravitational
potential of the companion star, and can then be
trapped in the potential of the compact star. Again, conservation of angular momentum generally
leads to spin-up. However, wind accretion is only important for stars with a strong stellar wind. These
are generally massive stars or stars in the final stages of evolution. Massive stars have a very short life
span leaving little time to spin up a compact object. Stars in the final stage of evolution will expand
to much larger radii, often also leading to Roche lobe overflow.
Low mass X-ray binaries
The term low mass indicates a low mass companion of the compact object. In low mass binaries the
companion evolves slowly, leaving much time for accretion. As stellar winds are negligible from low
mass stars, Roche lobe overflow is the main accretion mechanism in these systems. Compact objects
recycled in such a system thought to be the fast millisecond pulsars.
Low mass X-ray binaries are often transient binaries. In transients binaries, accretion occurs semiperiodically in outbursts, separated by periods of quiesence in which little or no accretion occurs. The
explanation of this behaviour is that accretion rates are faster than the rate of mass transfer. In such a
scenario, accretion starts when an instability develops in the accretion disk. The compact star accretes
matter faster than the companion transfers new matter to the disk. The accretion goes on until the disk
5
Chapter 1: Introduction
is completely accreted. During quiesence the disk is then slowly filled until a new instability generates
an outburst.
During the transient phase mass transfer will produce large amounts of X-rays, while also heating
up the surface of the compact star. During quiesence the compact star will cool down by emitting
X-rays. The cooling curve of such objects can be used to study the structure of the crust. This is
studied in the thesis of Degenaar (2006), where she concludes that the theoretical understanding of
crust cooling seems quite good, but needs further testing against observations. The theoretical results
do agree with current day observations but more accurate observations would allow for better testing.
1.1.3
Surface radiation of compact stars
Direct observation of radiation emitted from the compact star surface is currently possible using the
X-ray observatories circling Earth. This radiation can be used to study the compact star photosphere.
Using the spectrum, much can be learned about the precise structure of the crust and the atmosphere.
However, current day understanding of the spectrum is quite poor.
The emission profile also provides information on the cooling of the compact star. These cooling
properties can also be calculated from the equation of state and thus provide another mechanism,
besides mass-radius determination, to test equations of state. Such a calculation is done in Page et al.
(2006).
Neutron star cooling can occur by heat transport of neutrinos and photons. Let us first consider
neutrino cooling, as that is the most efficient mechanism in hot neutron stars. After fusing up to iron,
the core temperature of a star is usually over 1010 K. At these high temperatures, strong neutrino
emission will rapidly cool the star. Also when the star is much colder, neutrino emission will still be
the dominant cooling mechanism. Only for very low temperatures ∝ 106 K reached in about a million
years, photon cooling will become the dominant cooling mechanism. By calculating neutrino cooling,
and obtaining the heat capacity and conduction from the equation of state, it is possible to calculate
the thermal emission from compact stars as a function of their age (Page et al., 2006).
Since young compact stars often have their supernova remnants still around them, it is possible to
find the age of the compact star from the age of these remnants. These ages can be quite accurately
determined if the distance to the object is known. From the Doppler shift the velocity of the gas in the
supernova remnant is measured. From the distance and the angular size of the remnant an absolute
size of this remnant is determined. Using this information and the measured speed the age of the
remnant can be calculated.
Sometimes the age of a supernova can also be obtained from historic astronomical data. A famous example is the Crab pulsar, where according to Chinese astronomical archives a supernova has
occurred in 1054.
1.1.4
Free precession
Evidence from pulsar timing residuals indicate that long period free precession of the rotation axis
of a compact star may occur (Stairs et al., 2000). Free precession occurs as a reaction to a nonspherical disturbance of a rotating object. The precession period is proportional to the oblateness
of a compact star. However, it is also proportional to the amount of angular momentum stored in a
superfluid coupled to this precession (Shaham, 1977). This makes precession a good probe for the
interior properties of a compact star.
The usual model of superconductivity in a neutron star is one in which the superfluid and the
magnetic field are closely interlocked. As the crust is made of mostly iron, the crust is coupled
6
1.2 Formation and evolution of compact stars
strongly to the magnetic field as well. Superfluids have the peculiar properties under rotation, that lead
to strong damping of precession when coupled to the crust via the magnetic field. The occurrence of
long term precession indicates that this model may be flawed. Solutions to this problem are a different
model of superfluidity, or the occurrence of quark matter and colour superconductivity which has a
different coupling to magnetic fields. Superconductivity and colour superconductivity are reviewed
in chapter 6. In chapter 7 the consequences of electromagnetic and colour superconductivity on the
precession properties of a compact star will be studied.
1.2 Formation and evolution of compact stars
1.2.1
Stellar evolution and the formation of compact stars
Normal stars owe their stability due to the presence of a thermal gradient which balances the gravitational pull. The thermal gradient is maintained by nuclear fusion in the centre of the star. For about
ninety percent of its life, stars reside in the main sequence stage. In this stage burning hydrogen in
their core is the main source of energy, forming helium in the process. Our sun, for example, is in this
stage. A profound property of stars in the main sequence is that they are very stable. Major changes
in the characteristics of the star do not happen.
When a star runs out of hydrogen, helium burning has to set in. Since helium has a larger positive
charge, the electromagnetic forces between the nuclei will be much larger. In order to make it still
possible for the nuclei to come within the range of nuclear forces, a much larger temperature is needed.
To reach this temperature in the core, a steep temperature gradient is needed. To compensate this
temperature gradient, gravity must provide a compensating pressure gradient. This is possible only
for stars more massive than about 0.4 M .
If a star is lighter than this mass, a degenerate core of helium will form, while the outer layers will
be expelled by stellar winds creating a planetary nebula. The core will from a helium white dwarf.
Note that such light stars have extremely long lifetimes, longer than the current age of the universe.
Single star evolution therefore does not allow for the presence of helium white dwarfs in our current
universe. That they do exist is the result of binary evolution, in which mass transfer between the two
stars can seriously alter the evolution path of a star (see section 1.1.2).
Stars more massive than 0.4 M can burn helium. Lighter stars, with masses up to about 2 M ,
burn helium with degenerate cores where the temperature is low compared to density. In such a system, the average kinetic particle energy is much larger than the average thermal energy. Degeneracy
is discussed in more detail in chapter 4. In degenerate matter, pressure is not related to temperature.
This will lead to runaway nuclear burning: the helium will burn and heat up the star until degeneracy is lifted. This process is called the helium flash. Outside the helium core there is still hydrogen
present, leading to hydrogen-shell burning. Heavier stars can stably burn helium in their core. Also
these stars will burn hydrogen in a shell around their core. Later, unstable helium shell burning will
also occur, accompanied by large mass loss. The end products of helium burning are carbon, nitrogen
and oxygen.
Only stars more massive than M & 8M can use carbon, nitrogen and oxygen as fuel. These stars
have cores more massive than 1.4 M , which can overcome the electron degeneracy pressure and so
become dense enough to burn these elements. This mass is called the Chandrasekhar limit and is
caused by the failure of the electron degeneracy to effectively counter a strong gravitational pull. The
denser the star, the more closely packed the electrons must be. This will eventually cause the Fermi
energy, the energy at which the highest energy electrons reside, to go above the electron rest mass.
The electrons then become relativistic. Chandrasekhar showed that this leads to a maximum mass for
7
Chapter 1: Introduction
an object prevented from collapsing by electron degeneracy pressure. The outline of his derivation
can be found in 2.1.1.
Because the cores of massive stars can become hot enough to burn carbon, nitrogen and
oxygen, further fusion will occur. Again, shell
burning of lighter elements happens, often off
equilibrium. Heavy mass loss is common during
this phase. Almost always, stars of this mass are
able to burn all elements lighter than iron. However, no star can burn iron to heavier elements.
The reason is that iron has the most binding energy per nucleon (figure 1.5). Creating heavier
nuclei than iron does not release any energy. A
star with a core of iron will therefore have no
source of energy anymore.
Since the core is also too massive for electron degeneracy pressure to sustain balance, the
core will collapse beyond the white dwarf stage.
Figure 1.5: Binding energy per nucleon. As iron
has the maximum binding energy per nucleon, this
The core will collapse until a new mechanism
is the most stable element (image from , NASA)
is able to provide a balancing pressure. Neutron degeneracy is able to provide this pressure
against a much stronger gravitational pull (see Shapiro and Teukolsky, 1986). Also compact stars
made of free quarks seem to be able to provide the required pressure1 . Free quark matter is a state of
matter in which quarks behave as weakly interacting particles. In normal matter, quarks are strongly
bound in baryons as a result of confinement, the effect that no quark can isolated. However, at high
energies, the asymptotic freedom present in the theory of quark interactions indicates that deconfinement has to occur. Asymptotic freedom means that in the limit of infinite energies the interaction
strength goes to zero, as explained in section 3.3. The study of this quark pressure will be the main
subject of this thesis. Objects for which neutron or quark degeneracy pressure is the stabilising factor
will be very small, with radii of the order of ten kilometres. This is much smaller than typical radii
of white dwarf stars, which are or order 104 km, in which electron degeneracy pressure counters the
gravitational pull.
1.2.2
Supernovae
The collapse of a massive star creates a violent burst, a supernova. When a stellar core collapses,
the collapse releases huge amounts of gravitational energy. According to the virial theory, half of
this energy will be converted to heat. The massive heating this causes creates a shock wave, which
converts all iron in the core of the proto-compact star back to protons and neutrons. How this shock
wave is able to have so much energy to be able to do this conversion and also trigger the explosive
removal of the outer layers is currently not understood. Probably the neutrino flux caused by weak
processes and cooling of the superheated matter plays an important role. It can be calculated that
the free path length for neutrino in a very hot proto-neutron star is shorter than the radius of this star
(Reddy et al., 1998). This makes it possible for neutrinos to further heat up the proto-neutron star,
1
Publications include Alford et al. (2006a), Andersen and Strickland (2002), Buballa (2005), Fraga et al. (2001), Fraga
and Romatschke (2005), Fujihara et al. (2006), Glendenning (2000), Nicotra et al. (2006), Rajagopal and Wilczek (2000),
Rüster and Rischke (2004), Weber et al. (2006), Witten (1984)
8
1.2 Formation and evolution of compact stars
possibly driving the shock wave.
When more and more neutrons are formed due to beta decay, efficient cooling mechanisms become
unavailable as the main neutrino emission mechanism, direct Urca, is unable to conserve simultaneously energy and momentum. This causes the free path length of neutrinos to increase dramatically,
making the compact star virtually transparent for neutrinos. As calculated in Page et al. (2006) efficient neutrino cooling via the direct Urca process is only possible when the proton content of a
compact star is larger than ∼ 11 percent. This is discussed in more detail in section 1.2.4. There also
the direct Urca process is described.
It is also possible that the neutron degeneracy pressure or quark degeneracy pressure will not be
sufficient to counter gravity. In that case no known mechanism will resist the object from collapsing.
Then, the formation of a black hole is expected.
Supernova observations
Figure 1.6: A supernova in M51, a nearby galaxy. In the left frame an image of M51 before the supernova,
in the right frame the same galaxy after supernova 2005CS occurred. The bright dot in the middle is the
supernova. Supernova 2005CS has been identified as caused by an explosion of a star, probably similar to
supernova 1987A (photograph by GaBany, 2005)
Supernovae occur about three times per century in an average galaxy (Cappellaro et al., 1999), probably a similar rate occurs in our Galaxy. However, the latest certain supernova observation of a
supernova within our Galaxy dates from 1604. Although the explanation lies partially in statistics,
also the optical extinction due to dust in our galaxy blocking radiation from supernovae contributes.
Furthermore, Cassiopeia A is probably formed after this date but historic observations are not very
convincing. An explanation for this is that the luminosity of the supernova forming Cassiopeia A was
very low, as has recently been found in some supernovae in other galaxies as well (Tominaga et al.,
2005). Yet today, with the huge number of galaxies currently within visible range of our telescopes,
several tens of them are discovered each year.
On February 23, 1987 the first and at the time of writing this thesis only nearby supernova occurred
that could be studied with modern instruments. At that day, a blue giant star exploded in the Large
9
Chapter 1: Introduction
Magellanic Cloud, at a distance almost 170,000 light years. This supernova was visible to the naked
eye, and it was also the first object outside our solar system from which neutrinos have been identified
as belonging to this object. About three hours before the visible supernova explosion occurred, a 13
second neutrino burst was observed (see figure 1.7). This result indicates that it takes about three hours
for the shock wave to propagate to the surface of the star, matching model calculations. Furthermore,
this event allowed for accurate testing of some neutrino properties. In the remnant of supernova
1987A, as this event is called, no compact star has yet been discovered. It is expected that in due time
the compact star will be found, as these neutrino observations are in agreement with cooling models
of a proto-compact star.
In other supernova remnants compact stars
have been discovered. Both the Crab nebula
(see figure 1.1) and Cassiopeia A host a compact star. The Crab nebula is associated with
a supernova that went off in 1054 and was observed by Chinese astronomers. Cassiopeia A
probably went supernova around 1680, although
no clear recording was made in that year. There
Figure 1.7: The detection of neutrino’s by the
is a record of the appearance of a new star in
Japanese Kamiokande II and the IMB experiment
1667, which may be the supernova related to
(image from UNO collaboration, 2001)
Cassiopeia A.
1.2.3
Magnetic fields
One of the most important characteristics of a compact star is its magnetic field. Ranging from 104
T for millisecond pulsars up to 1012 T for magnetars, these strong magnetic fields have an important
influence on the compact star. They are the main cause of pulsar emission, as discussed in section
1.1.1. In binary systems the magnetic field of the compact star will interact with the accretion disk.
This can lead to very complex behaviour of accretion and may be related to some of the quasi periodic
signals the objects emit (Boutloukos et al., 2006). It also tunnels the accretion to the magnetic poles,
generating hotspots on these poles. The X-ray production occurs in these hotspots leading to pulsed
X-ray emission with the frequency of the neutron star its spin.
Magnetic fields are thought to be present as a result of magnetic flux conservation. Massive
stars, capable of forming a compact star, have magnetic fields present. These fields are normal stellar
magnetic fields, thus much weaker than compact star fields. During the collapse the magnetic flux is
preserved. As a result of the order 106 decrease of the radius the magnetic field is boosted by an order
of 1012 . Assuming a field similar to our sun, this would lead to a 109 T field. These fields perfectly
match the fields of normal compact stars.
Recently it has been discovered that there probably is a significant difference in magnetic fields of
the stars that form compact stars, the B stars2 (Ferrario and Wickramasinghe, 2005). There exist high
field B stars, having fields of an order 100 stronger than normal B stars. It is thought that these stars
might evolve to magnetars when they collapse into a compact star. This story is similar to strong field
white dwarfs, which make up a similar fraction of the total number of white dwarfs as the fraction of
strong field white dwarf forming A stars to normal A stars, the stars that create these white dwarfs.
This suggests that the strong field stars evolve into strong field white dwarfs.
2
The letter represents the spectral type of the star, which is related to its surface temperature. B stars have a surface
temperature ∝ 2 × 104 K, A stars ∝ 1 × 104 K. The complete list from hot to cold is O B A F G K M, the sun is of type G.
10
1.2 Formation and evolution of compact stars
An observational challenge is provided by the low number of massive stars. Massive stars are
much more seldomly produced, as found by Salpeter based on a statistical analysis of the local stellar
population (Salpeter, 1955). The fitted power law relation is
M0 ∝ M −3.5
(1.3)
Also, they are very short lived, with a life span ∝ 10 million years. As a result massive stars are not a
frequent occurrence, making statistics much more difficult. Another problem is the so called selection
effect, that causes high field compact stars to be found much more easily. Due to these problems the
situation for compact stars is not very clear cut.
Another possibility for creating the strong field compact stars, is by invoking a dynamo action
during its collapse. Although not known how this should work, it is not that strange an idea that the
massive friction during a collapse could in principle create huge magnetic fields. A drawback from
this view is that it is not clear why in some supernovae this dynamo action would be present, while in
others it has to be absent to predict the normal compact stars.
For the millisecond pulsars another mechanism has to be invoked. These compact stars are characterised by a, for compact stars, weak field of about 104 T. The magnetic breaking as a consequence
of its field is such that the time scale for a significant change in the rotation period is larger than
the Hubble time. Therefore, it can be assumed that the periods at which these stars rotate haven’t
significantly changed during their lifetimes.
As discussed earlier, millisecond pulsars are probably generated by accretion in binary systems.
How accretion leads to weak fields is still subject of much debate. One possibility is that the compact
star first spins down to a very slow period. However, this result is contradicted by observations of
slowly rotating compact stars in binary systems which still posses normal magnetic fields ∼ 108 T
(Coburn et al., 2006). Another possibility is that the magnetic field is buried by accretion. It is then
still present, but it doesn’t emerge from the surface and is therefore not observable (Bhattacharya
and Srinivasan, 1995). None of the models is currently able to provide an established mechanism to
explain the weak fields of millisecond pulsars, which seems to cluster at around 104 T (see figure 1.3).
Evolution of magnetic fields in isolated pulsars
As can be seen in figure 1.3 the lifetime of a pulsar strongly depends on the strength of the magnetic field. High-field pulsars are expected to produce pulsar-radiation for only about ten million
years, while the low field millisecond pulsars can in principle have existed since the formation of the
universe, as their life span based on spindown timescales is longer than the Hubble time.
After the discovery of a few dozen pulsar systems, there seemed to be evidence for the decay
of the magnetic field of the high-field pulsars with a decay time similar to its lifetime, ∝ 10 million
years. After the discovery of many more pulsars the evidence for the decay of the magnetic field has
diminished, leading to the opinion that these fields are stable for much longer than the pulsar lifetime
(Bhattacharya and Srinivasan, 1995). The pulsar lifetime is defined as the time pulsars are capable of
producing electromagnetic radiation by the pulsar mechanism. As seen in figure 1.3 it seems that if
pulsars rotate too slow this mechanism no longer works. For pulsars with a normal field ∼ 108 T this
occurs for rotation periods of about 10 seconds.
As discussed in section 1.1.1 one of the possibilities for the stability of the magnetic field is the
presence of a superconducting medium that is coupled to the magnetic field. Such a system would
only show a very slow decay of the field, well within observational bounds. A detailed analysis is
provided in section 7.1. Yet, the superconducting picture is not necessary, as degenerate matter is a
very good conductor anyway. For neutron matter this conductivity is about 1028 s−1 , leading to decay
11
Chapter 1: Introduction
times ∼ 1012 yr, larger than the Hubble time (Baym et al., 1969). For quark matter it can be calculated
that Ohmic decay times are of similar order ∼ 1013 yr (Alford et al., 2000). Ohmic decay times in the
crust are much shorter (Cumming et al., 2004). See Harding and Lai (2006) for a recent discussion.
Field decay in binary systems
In contrary to solitary systems, fields of compact stars in binary systems do decay. Evidence for this
decay is provided by observations of X-ray binaries. In high-mass X-ray binaries, the compact star
is often pulsating, suggesting a strong field able to generate pulsed X-ray emission. In contrast, the
compact star in low-mass X-ray binaries is seldom a pulsating object, an indication for much weaker
fields. Since the high-mass binaries are generally much younger than the low-mass binaries, this can
be seen as an indication of field decay. This indication is strengthened further by the low fields of
millisecond pulsars, for which there is good evidence that they originate from the low-mass X-ray
binaries (Bhattacharya and Srinivasan, 1995).
However, as these object undergo mass transfer there might be other mechanisms at work altering
their magnetic field. Mass transfer severely heats up the crust, and also alters its equation of state.
This might lead to an extra Ohmic decay component. Yet, it is unclear how such a mechanism is
capable of reducing the core magnetic field. Suggestions are that these fields are somehow buried,
although the mechanisms involved are still very unclear.
Magnetic field evolution of magnetars
Due to the extremely high magnetic fields of magnetars (B ∼ 1010 − 1012 T) the evolution of their
magnetic fields is different. Part of the radiation emitted by these objects is probably generated by the
decay of the field, as the total energy emitted is larger than the observed spin-down.
If the matter inside the object is of hadronic origin, the field probably exceeds the critical field
for superconductivity. This might present a mechanism for Ohmic dissipation. Also reconnecting
field lines can liberate much energy. Such events are thought to be the origin of giant outbursts as are
observed in the soft gamma ray repeaters (SGR). How Ohmic decay can be strong enough to explain
the decay is not clear.
If the interior is made up of quark matter the critical field of colour superconductivity will not be
reached (see section 6.2, Alford et al. 2000), so there is no comparable mechanism available for field
decay in compact stars made of quark matter.
1.2.4
Cooling of compact stars
The age of a compact star is important, as knowledge of the age allows for testing evolutionary models
of the compact star. When a compact star is born, it is initially very hot, with temperatures larger
than 1010 K. This large temperature is the result both from the supernova as well as the initial core
temperature, which must be of order 1010 K to allow burning up to iron in the first place. At these
temperatures, nuclei become unstable and can be thermally destroyed. This results in the breakup of
all nuclei to protons and neutrons. The degeneracy pressure of the electrons then makes it energetically
favourable to convert electrons and protons into neutrons via inverse beta decay
p + e− → n + νe
(1.4)
resulting in the production of a neutrino flux. The reverse of this process, beta decay, is also possible:
n → p + e− + νe . The two processes combined are able to convert heat into neutrinos at a very fast
12
1.2 Formation and evolution of compact stars
pace. This is called the Urca process, after a Brazilian casino where legend says that it is possible to
loose money at an equally fast rate.
q
This neutrino flux quickly cools the neutron star to
temperatures ∝ 109 K, when the inverse beta decay pro0
Z
cess becomes unavailable due to the degeneracy of the
ν neutrons blocking momenta of the final state fermions. In
order for the direct Urca process to occur the momenta of
participating particles must lie close to the Fermi surface.
ν̄ Lattimer et al. (1991) have shown that this condition is
met only when the proton fraction exceeds 11%.
Below this threshold stage, modified Urca processes
q
become the dominant cooling mechanisms. This process
Figure 1.8: Production of neutrino
requires an additional spectator baryon to carry away the
bremsstrahlung in a quark scattering dimomentum. As this is a three particle process, the rate
agram
will be much slower. Other processes cooling a neutron
star at this stage include neutrino-bremsstrahlung (Price, 1980). In this process, a baryon is accelerated
via the nuclear interaction and a neutral current Z0 is produced. The Z0 then decays very quickly to two
neutrinos. This process is very similar to the electromagnetic bremsstrahlung in particle accelerators,
except of course that the so produced photons are a stable final state.
It is also possible that in the compact star free quark matter will be present. This leads to a
somewhat different cooling behaviour. However, since quarks also carry isospin charge, the main
characteristics are the same. Quarks also undergo direct or modified Urca cooling, and also quarks
produce neutrino bremsstrahlung. The difference is that in a state in which free quark matter is present
there are more particles. More particles generally leads to faster cooling, and thus the expectation that
quark stars are somewhat cooler than stars consisting only of neutrons.
The story above is complicated by the presence of superfluid states. Both quarks and neutrons
will probably undergo a superfluid transition at temperatures ∼ 109 K for neutrons and ∼ 1010 K for
quarks. As superfluidity causes a gap between the superfluid ground state and excitations, final states
for scattering become even less available. The critical temperature for quark matter is larger than the
critical temperature for neutron matter. This may actually cause quark stars to be hotter than their
neutron counterparts. Note that the critical temperature is of the order of the formation temperature
of quark stars. These objects might thus be in a superconducting state immediately after formation.
Colour superconductivity as present in weakly interacting quark matter can have important influence on the quark equation of state. In chapter 6 superconductivity and colour superconductivity will
be explored further. Cooling models of compact stars are studied in Page et al. (2006).
13
CHAPTER 2
Structure of compact stars
In this chapter the general structure of a compact star is reviewed. First we will derive the general
equations of stellar structure. These equations must be satisfied by any model describing a spherical
star. In order to solve the models, we will need an equation of state, relating pressure and density.
Using a power law equation of state, a so-called polytropic relation, the general properties of stellar
models can be studied. We will derive why a polytropic relation is useful and study its behaviour.
In the second part of this chapter we will look at some complications to the picture sketched in the
first part. More general equations of state will be considered, and the effect of rotation on the structure
equations is studied.
2.1 General structure equations
The structure of a compact object is determined by its equation of state and the equations for hydrostatic equilibrium. In this thesis, we will deal with massive objects with small radii, so taking the
weak field limit for the gravitational potential is not appropriate. Therefore, we cannot use Newtonian
formulae, but have to use a general relativistic formulation.
For a spherical, non-rotating star in equilibrium the relativistic hydrostatic equation was first derived by Tolman (1939) and Oppenheimer and Volkoff (1939). It can be derived from the Schwarzschild metric using a nonzero energy momentum tensor (see appendix A for the derivation). The TOV
equation reads
dP G( + P)(m(r)c2 + 4πr3 P)
= 2 4
dr
r c (1 − 2Gm(r)/(c2 r))
(2.1)
In this equation c is kept explicitly, in order to understand the relation with Newtonian mechanics.
The pressure P and the energy density are both given in units of energy per volume. The enclosed
mass m(r) is given by equation
dm(r)
r2
= 4π 2
dr
c
(2.2)
with again the energy density. Note that as c → ∞, the TOV equation asymptotically approaches the
equation of hydrostatic equilibrium based on a Newtonian gravitational potential, as it should. Then,
15
Chapter 2: Structure of compact stars
in equation 2.2 the energy density has to be replaced by the matter density ρ.
dP Gm(r)ρ
=
dr
r2
dm(r)
= 4πr2 ρ
dr
(2.3)
(2.4)
Note that M(r) defined by
M(R) =
R
Z
0
dm
dr
dr
(2.5)
is the total gravitational mass. Generally, the gravitational mass of a compact star is reduced with
respect to the integrated energy density due to its gravitational binding energy. The difference between
gravitational mass and integrated energy density can be quite significant. This difference is taken into
account by the metric part of the Einstein equation. More on this can be found in appendix A.
The equations of hydrostatic equilibrium must be supplemented by an equation of state to form a
closed set of equations. The equation of state is non-trivial and requires knowledge of the composition
of the object. In case of a perfect fermion gas an analytic solution to the equation of state is possible.
In the limit of non-relativistic or extreme relativistic degenerate gases the equation of state reduces to
a power law dependence of pressure P and density ρ. This is know as a polytropic equation
P = Kργ
(2.6)
with K a constant and γ the power of the polytrope. For non-relativistic polytropes γ = 5/3, while
γ = 4/3 for relativistic polytropes.
2.1.1
Polytropes
The polytropic law can is obtained using an ideal degenerate gas. Such a gas has an energy density
(p) =
q
m2 + p2F
(2.7)
where pF is defined as the Fermi momentum. Taking x ≡ pF /m in this equation, we get
q
(zF ) = m[ 1 + z2F − 1]
These equations can be used to write down the number density
Z zF
4πg(z) 3
z
n = g(pF )m−3
4πz2 dz =
3m3
0
(2.8)
(2.9)
Here g(pF ) is the number of states per Fermi level. For spin-1/2 particles g(pF ) = 2. Similarly, the
pressure is given by
Z zF
z4 dz
4
(2.10)
P = g(pF )m
(1 + z2 )1/2
0
This equation is conveniently written as A f (z), where A is a dimensionful constant while the dimensionless function f (z) is
1
f (z) = z(2z2 − 3)(1 + z2 ) /2 + 3 sinh−1 (z)
(2.11)
16
2.1 General structure equations
In the limit of small or large z we obtain from this equation the limits:
P∝
µ
!5/3
P∝
µ
!4/3
z1
(2.12)
z1
(2.13)
A careful derivation of the equation of state for an ideal degenerate gas is provided in section 4.1.
Together with the equations for hydrostatic equilibrium and mass conservation this equation can
be solved to obtain the mass and radius of a star as a function of central pressure. For non-relativistic
degenerate gases, the radius will decrease with increasing mass, while for the relativistic case the
dependence on central pressure drops out and a maximum mass appears, as will be shown below.
When using the Newtonian equation of hydrostatic equilibrium and a polytropic equation of state,
there exists an exact solution. Combining equations 2.3 and 2.2 yields
!
1 d r2 dP
= −4πGρ
(2.14)
r2 dr ρ dr
while the polytropic equation of state (equation 2.6) can be written in dimensionless form
s
a=
ρ = ρc θ n
(2.15)
r = aξ
(2.16)
(n + 1)Kρc/n−1
4πG
(2.17)
1
with ρc = ρ(r = 0) and n defined as
1
n
Using the dimensionless equations 2.15-2.17 to rewrite equation 2.14 one obtains
γ =1+
1 d 2 dθ
ξ
= −θn
ξ2 dξ dξ
(2.18)
(2.19)
This equation is called the Lane-Emden equation. The boundary conditions that close the set of
equations can be obtained by requiring that ρ(0) = ρc and ρ(r) = 0. These conditions are θ(0) = 1 and
θ(1) = 0.
Its use to dense structures is that it gives an upper bound to stellar masses and radii, as general
relativity acts as an extra attractive interaction at high densities. The functions for mass and radius
obtained from the Lane-Emden equation are
M = 4πa
which combine to
(3−n)/(1−n)
M = 4πR
"
3
R = aξ1
(2.20)
ρc ξ12 |θ0 (ξ1 )|
(2.21)
(n + 1)K
4πG
#n/(n−1)
(3−n)/(1−n)
ξ1
ξ12 |θ0 (ξ1 )|
(2.22)
where ξ1 is defined by θ(ξ1 ) = 0, which can be solved numerically from equation 2.14.
17
Chapter 2: Structure of compact stars
For a relativistic polytrope (γ = 4/3) the solution for M is independent of the central density. Using
the numerical results ξ1 = 6.89685 and ξ12 |θ0 (ξ1 )| (Shapiro and Teukolsky, 1986)
2
M = 1.457
µe
!2
M
(2.23)
with µe is the mean molecular weight. This limit can also be obtained by an energy balance between
gravitational binding energy Egrav and the internal energy of the relativistic Fermi gas Eint (Shapiro
and Teukolsky, 1986)
N 1/3 GNm2b
E = Eint + Egrav =
−
=0
(2.24)
R
R
which yields a similar mass as equation 2.23.
The advantage of this equation is that it is easy to add corrections to internal energy and the
gravitational potential. In this manner, it is possible to correct equation 2.24 in such a way that general
relativity is accounted for. The correction factor is calculated from the TOV equation in Shapiro and
Teukolsky (1986) and is
2
7
∆EGR = −0.918294M /3 ρc/3
(2.25)
which enhances gravitational binding, as expected. Note that this factor is no longer correct when
other corrections to equation 2.24 are made. It will, however, still be true that general relativity
deepens the gravitational potential well.
Taking into account other interactions than gravity is also possible. This is important, as an ideal
gas assumption works quite well to describe white dwarfs, but models based on an ideal gas for
neutron and quark stars do not agree with observations. This can be understood by realising that
interactions between neutrons and quarks cannot be considered small, so an ideal gas approach is not
adequate. A simple solution is to express the mean field interactions as a vacuum pressure and include
this term. This will be discussed in section 5.1.1.
2.2 Modelling stars
2.2.1
Models of neutron stars and quark stars
There are typically two distinct models of neutron and quark stars, and a number of adaptations to this.
A schematic overview of these models is given in figure 2.1. The oldest model is the pure neutron
matter model. This model consists of a crystalline iron crust, which becomes more neutron rich deeper
in the star. At a certain depth the neutrons will drip out, forming a neutron fluid. This fluid is probably
a superfluid, a fluid that can flow without friction but also has some peculiar properties when rotated.
When rotating a superfluid, not the superfluid itself will rotate, but all angular momentum will be
contained in small vortices where the matter is not in a superfluid state. This is a result of the broken
rotational symmetry due to long-range order in the superfluid state.
The vortices will pin to impurities as that results in a state of lowest energy. In a neutron star, these
impurities are provided by the crust lattice. Movement of pinned vortices can only occur by vortex
creep, a process in which vortices tunnel through the energy barriers created by pinning (Anderson,
1962). Pinning to the lattice and an inability to have a vortex creep in a certain region of the star can
be used to explain glitches (section 1.1.1). Vortex pinning and its consequences for glitches will be
further explored in section 7.2.1.
Deeper in the neutron star atomic nuclei become unstable, so also free protons appear. These protons are thought to form a superconducting fluid. The appearance of superconductivity has important
18
2.2 Modelling stars
influences on the behaviour of the magnetic field of the neutron star, which will be the subject of
chapter 7.
In the neutron star centre, very high densities may lead to deconfinement. Then, the dense interior
must be described using a quark equation of state. We will call such a star a hybrid star. It is also
possible that other forms of matter may be present in the interior, like strange quark containing baryons
and mesons (Hyperons and Kaons) as well as drops of strange matter.
Another interesting possibility is that when strange matter is created the energy released by the
conversion may actually trigger the whole star to convert to strange matter (Witten, 1984). This object
is called a quark star or strange star. These objects may be visible as bare quark stars, but it is more
likely that they will have a very thin crust. Such a crust would appear if charge neutrality is not present
in the quark matter. This is likely due to the mass difference of the up, down and strange quarks. In
this case, electrons must be present for charge neutrality. At the surface, the quark density will drop to
zero over a very short distance of a few femtometer. This is much faster than the electric field falls off,
leading to a strong electric field. Such a field would be able to support a crust with a mass ∝ 10−5 M
(Alford et al., 2006b, Jaikumar et al., 2006b, Stejner and Madsen, 2005). In this case the quark star
will probably have an emission very similar to that of a neutron star.
2.2.2
Solving the models
The equations of state and the structure equations 2.1 and 2.2 yield, at zero temperature, a closed
set of equations given two boundary conditions. For nonzero temperatures equations to describe heat
loss and energy production in the star are also needed. The dependence on temperature is of minor
importance for both neutron stars and all conceivable quark star models, since the interaction energy
is generally much higher than the thermal energy available in such stars. It is therefore safe to assume
zero temperature when determining the structure of compact stars.
Analytically solving this set of equations is in general not possible. For a polytropic equation
of state there do exist exact solutions (see for example Shapiro and Teukolsky, 1986), but for a more
realistic one numerical solutions have to be used. Solving the set of equations is done by first assuming
an arbitrary central pressure as the first boundary condition. Using the structure equations one can
then find a new pressure a distance δx from the origin. Repeating this process until the surface is
reached yields a model of the star. The location of the surface is determined by a boundary condition,
generally the surface is assumed to exist at the radius where the pressure vanishes. A detailed analysis
is provided in chapter 5.
In the structure equations 2.1 and 2.2 a non-rotating, spherical star is assumed. Of course a more
general set of equations can be used, making the problem to be solved more complex and thus more
expensive in computing time. However, if these effects are of great influence, they should not be
neglected. The effects of rotation are studied in section 2.2.4, and are found to be of some importance
for the millisecond pulsars. However, the effects of rotation will only lead to small corrections and
can be neglected as the errors resulting from uncertainties in the equation of state are much larger.
Due to the extreme gravitational pull of a compact object, the structure will be completely determined by the shape of the potential well. In case of a non-rotating star, a spherical potential is the
state of minimum energy. Since it can be assumed that surfaces of equal pressure match equipotential
surfaces in such strong potentials, the star needs to have a spherical shape. The assumption that the
potential determines the shape means that gravity is much stronger than long-ranging structural forces
that can exist in materials. As this is already the case for planets, it is certainly the case for compact
objects, so a spherical potential is a good assumption.
19
Chapter 2: Structure of compact stars
2.2.3
Equations of state
Figure 2.1: Possible model of a compact star. The radii of the different components may vary, depending
on the model used and the object mass. Not shown is the structure that might be present in between the
crust and proton-neutron fluid. Here a complicated area might be present, were fore example vortices of
the neutron superfluid can interact with the crystal lattice structure of the iron crust.
In section 2.1 the equation of state of a degenerate ideal gas was studied. We found that in the limiting
case of relativistic or non-relativistic equations, simple polytropic behaviour occurs. In intermediate
cases, a linear combination of these limiting cases will apply.
However, matter found in compact objects is not an ideal gas. While the Coulomb forces in
white dwarf stars may be weak and therefore considered to be a minor perturbation to the ideal gas
approximation, in denser neutron stars and quark stars the interparticle forces are much stronger. This
leads to a serious complication of the problem of finding an equation of state. Another problem is that
the behaviour of dense matter is poorly known, as it is impossible to study such matter on Earth. In
experiments at GSI in Darmstadt and RHIC in Brookhaven heavy ions are collided in order to obtain
high densities, but these experiments are unable to obtain supernuclear densities at low temperatures
(Buballa, 2005). The ALICE experiment at the LHC in Geneva will suffer from the same limitations.
The result is that there is no equation of state for neutron matter that is generally accepted.
The strongest force in neutron stars and quark stars is the strong force. A major characteristic
of the strong force is asymptotic freedom as described in section 3.3. It might be that due to the
high densities inside neutron stars the strong coupling constant becomes so small that deconfinement
occurs. Although deconfinement cannot be observed at high densities and low temperatures, at high
temperatures it can be observed in particle accelerators. There is evidence that in RHIC the quarkgluon plasma, as the high temperature state is called, has been observed (Gyulassy and McLerran,
2005), although the result is still controversial (Star Collaboration, 2005). Also in the early universe
this quark-gluon plasma must have been present.
While the strong force will be studied in chapter 3, it is interesting to consider the qualitative
20
2.2 Modelling stars
consequences of this possibility. It is important to realise that quarks are light particles. The Fermi
energy of quarks in a baryon is about 200 MeV, which is already much larger than the likely mass of
up- and down quarks (both around 5 MeV). This energy is estimated using a baryon size of 1 fm. The
uncertainty principle requires momentum and energy to be at least ~. Using that ~c ' 200 MeVfm,
the uncertainty principle indicates that a particle in a 1 fm box needs an energy of at least 200 MeV.
Since this energy is much larger than the rest energy, it is safe to consider these particles massless
in equations. The strange quark is more massive (about 100 MeV), but with typical Fermi momenta
achieved at high densities (at least 200 MeV) this particle will also become relativistic. However,
ignoring the strange quark mass does not seem justified as will be shown in chapter 5
For a model of relativistic particles to provide stars with a range of masses, it is vitally important
to include their interactions. Otherwise the result will be a relativistic ideal gas, which allows one
solution only, as shown in 2.1. The equation of state of weakly interacting quark matter leads to a
completely different behaviour of the mass-radius relation. At lower masses, quark stars will grow if
the star becomes more massive, indicating that quark matter behaves as an incompressible medium.
Only near the maximum mass a quark star can have an increase in mass leads to a smaller radius, just
as neutron stars. The details of the quark star equation of state will be studied in chapter 4 and 5.
2.2.4
Effect of rotation on compact star structure
Some compact stars are known to spin very rapidly (see section 1.1.1). In this case, it is important to
realise that centrifugal forces might be an important influence to the structure. Obtaining an equation
for the potential can be achieved by solving the Kerr metric with an ideal fluid energy-momentum
tensor. However, this ‘Kerr-TOV’ equation has currently not been found, leaving only numerical
methods to calculate the potential of a spinning compact star.
In Colpi and Miller (1992) it is shown that fast rotation can have profound effect on the massradius relation, leading to larger radii (up to 0.3 larger) for sources rotating at near breakup speeds
compared to non-rotating objects. Furthermore, fast periods allow a somewhat higher maximum mass
(about 20%). The calculations in Colpi and Miller (1992) were done assuming a star rotating at
near breakup speeds. In this section we will estimate the effect of rotation on more slowly spinning
compact stars found in nature. We will see that spin can generally considered to be slow, even for the
fastest spinning pulsars.
An estimate of the importance of spin can be made by a comparison of the centrifugal force Fc to
the gravitational force FG , with the forces given by (Colpi and Miller, 1992)
Fc = rΩ2
1 − 3M0 /r
1 − 2M0 /r − r2 Ω2
FG = M0 /R2
(2.26)
(2.27)
Fc is the general relativistic equation for a particle to move around an object of mass M0 and distance
r from the centre. In this equation G = c = 1, and the gravitational force is expressed by a Newtonian
equation. Although the weak-field limit is not appropriate for compact stars, the effect of spin will
prove to be so small that a careful analysis is not necessary.
Calculating Fc /FG as a function of Ω and setting it equal to 1, gives Ω ≈ 2000 s−1 , for a 1.4 M
compact star with a radius R = 11 km. So a pulsar has to spin at 2000 Hz to reach breakup speed.
For a long time the fastest known pulsar rotated at a frequency of 642 Hz (1.6 ms). Recently faster
rotating pulsars have been found. In March 2006 a pulsar rotating at 716 Hz (1.4 ms) was claimed
(Hessels et al., 2006), which is currently the fastest spinning pulsar known. For a period of 1.4 ms
the fraction Fc /FG ' 0.08. As this fraction is smaller for slower rotating pulsars, the assumption that
21
Chapter 2: Structure of compact stars
a slow rotation approximation can be used seems justified. Yet, these results only apply for objects
rotating at near breakup velocities.
In addition to this, the uncertainties in the equation of state of neutron stars and quark stars are
larger than a possible alteration due to rotation. Again, thus suggests it is safe to ignore the consequences of rotation. Therefore, throughout this thesis rotation is ignored.
22
CHAPTER 3
Quantum chromodynamics at high
densities
Quarks are charged with respect to all forces. However, the theory of quantum chromodynamics
has by far the strongest coupling constant, thereby being of greatest influence on the equation of
state of the quarks. Weak nuclear interactions are essential to describe beta equilibrium and cooling
mechanisms, electromagnetism is assumed to be of only minor importance. In this chapter we will
review some of the quantum field theory needed to understand quarks at high densities.
3.1 Symmetry groups
Symmetries of systems are almost always important. Neuther’s theorem states that symmetries are
coupled to conserved quantities. Examples are the translational symmetry giving rise to momentum
conservation and gauge symmetry leading to charge conservation.
In order to understand the properties of quark matter, it will be useful to use the symmetries of
the system. The symmetries discussed are mostly abstract, mathematical symmetries. There will be
one exception, however, and that is the symmetry of rotation. Rotational symmetry is broken by a
superfluid state (see chapter 7) which may also occur in quark matter.
Rotational symmetry is a fundamental property of unordered isotropic systems. It is best described
by stating that some function describing the system should be unaltered by rotations. A useful way
is to consider the set of all rotations as a group G. This set can be described as follows: a rotation is
a transformation of a system that preserves the inner product of two vectors. The set of all operators
satisfying this condition in a three dimensional space is called SO(3), the group of special orthogonal
3 × 3 matrices. Special states that det(M) = 1 ∀M ∈ G, orthogonal means that M −1 = M T .
It is possible to reproduce all the necessary matrices from just three generators. These generators
are in essence infinitesimal rotations along the x, y and z axis. The procedure to obtain a group element
is to take the exponent of the generators M = exp[iθn̂ · L] with n̂ the rotation axis and L the generators
of SO(3). In the adjoint representation the generators are


 0 0 0 


L x =  0 0 −i  ,


0 i 0


 0 0 i 


Ly =  0 0 0  ,


−i 0 0


 0 −i 0 


Ly =  i 0 0 


0 0 0
where φ x,y,z are the angles of rotation along the x, y and z axis, respectively.
23
Chapter 3: Quantum chromodynamics at high densities
In a similar fashion we can now also introduce new symmetries. We would like to have a unitary
theory so that preservation laws are obeyed, thus we better use unitary operators. Unitary matrices
have the property that the hermitian conjugate is equal to the inverse M −1 = M † . The simplest example
of such a group is U(1) defined as
M = eiφ
(3.1)
Expectation values are globally symmetric under this group, as
hψ|ψi → hψeiφ |ψeiφ i = hψ| e−iφ eiφ |ψi = hψ|ψi
(3.2)
We will see later in this chapter that, if we require the symmetry of equation 3.1 to become a local
symmetry, which means that φ depends on x, we need to alter our theory to include electromagnetic
interactions.
Of more general unitary symmetry groups we only need to consider det(M) = 1, as all unitary
matrices with det(M) , 1 are just a product of a special unitary matrix with an element of U(1). We
choose to consider U(1) separately. The simplest nontrivial example then is SU(2). It is defined as all
unitary matrices of the form
!
a b
M=
c d
with ad − bc = 1. Note that in general two elements of this group do not commute: it is a non-Abelian
group. It is generated by the Pauli matrices
!
!
!
0 1
0 i
1 0
σ1 =
σ2 =
σ3 =
1 0
i 0
0 −1
It can be shown that SU(2) is locally isomorphic to SO(3), the group of rotations. This explains why
angular momentum and spin can be added in atomic physics as if it were the same property. However,
the SU(2) symmetry can also be seen as the symmetry of the isospin interaction.
The next relevant group is SU(3), the symmetry group of the strong interactions. It is generated
by the eight Gell-Mann matrices, therefore this symmetry will need eight gauge bosons: the eight
gluons. Again this is a non-Abelian group. This will lead to charged gauge bosons, causing asymptotic
freedom as discussed in 3.3.
3.2 Quantum field theory
Combining quantum mechanics with special relativity requires the concept of a field. In special relativity, observers do not need to agree on observables such as energy. Since this theory also makes
it possible to convert energy to particles via E = mc2 , a fixed-number particle theory does not work.
Needed is a theory describing in principle infinitely many particles, and the most natural choice is
then a field theory. In quantum field theory wave functions describing particles become operators,
creating particles from the vacuum. In a free theory this leads to amazingly simple equations, such as
the Klein-Gordon equation for a free scalar field φ(x)
(∂µ ∂µ + m2 )φ(x) = 0
(3.3)
or the Dirac equation for a free fermion field ψ(x)
(i∂/ − m)ψ(x) = 0
24
(3.4)
3.2 Quantum field theory
As usual, the summation over multiply occurring indices is implied. When quantised, these equations
result in creation and annihilation operators, identified as the operators creating and annihilating the
particles. The negative energies occurring in the equation can be interpreted as antiparticles.
These equations possess global symmetries, like the U(1) discussed in section 3.1. Requiring these
symmetries to be also locally valid, it is possible to acquire an interacting theory. For example the
U(1) symmetry can be used to derive electromagnetism. Explicitly for an arbitrary symmetry group
φ(x) → eigM(x) φ(x)
φ (x) → e
∗
∂µ φ(x) → e
(3.5)
φ (x)
−igM(x) ∗
igM(x)
∂µ φ(x) + i∂µ M(x)e
(3.6)
iM(x)
φ(x)
(3.7)
were M(x) is an element from a gauge group, and g is the coupling strength. The term containing the
derivative breaks gauge invariance. This can be solved by introducing a covariant derivative which
does transform in the desired way
Dµ φ(x) → eigM(x) Dµ φ(x)
(3.8)
Dµ ≡ ∂µ + igAµ
(3.9)
Aµ → Aµ − ∂µ M
(3.10)
This can be done by requiring
with transformations of Aµ satisfying
This last requirement is exactly the gauge invariance present in massless vector fields, like photon
and gluon fields. Introducing this derivative in the free Lagrangian density L
1
L = ψ(i∂/ − M)ψ → L = ψ(i∂/ − M)ψ + gψγµ ψAµ + F µν Fµν
4
(3.11)
leads to an interaction Lagrangian via the interaction term gψ̄γµ ψAµ . Finally, to describe the gauge
field behaviour, inclusion of
Fµν = ∂µ Aν − ∂ν Aµ + ig[Aµ , Aν ]
(3.12)
is necessary.
For M an element of U(1), the counterterms Aµ can be identified as gauge bosons of the electromagnetic field, and also the interaction between particle and photon is present. The commutator in
3.12 vanishes for an Abelian symmetry group like U(1).
In a similar way, SU(2) leads to the weak interactions and SU(3) to the strong interactions. Complication to the latter two theories is that they are formed by non-Abelian symmetry groups (not all
elements of the group G commute: ∃A, B ∈ G : AB , BA). In this case the commutator in 3.12 cannot
be removed from the Lagrangian, leading to interactions between the gauge bosons.
For realistic quantum field theories, no analytic solutions are available1 . However, it is possible
to write the solution as a series of Green’s functions, yielding a perturbation series expansion. Each
next order in this series expansion will be of a higher power in the coupling constant. With a small
coupling constant higher order terms will soon provide only minor corrections.
The most convenient way to find this series expansion is by using Feynman diagrams. Feynman
showed that with only a small number of rules, the Feynman rules, it is possible to write down all
1
Only a few interacting field theories can be solved analytically, and only in two dimensions (see Peskin and Schroeder,
1995, ch. 22, and references therein)
25
Chapter 3: Quantum chromodynamics at high densities
occurring terms. Connecting the external lines in all topological different ways using a maximum of
n vertices, n being the order of your expansion, you find all possible amplitudes from the in-state to
the out-state at that order. By this method an in principle difficult mathematical problem has been
reduced to finding all different diagrams and calculating the absolute square of the sum of diagrams.
This is still not an easy task, but at least calculations are possible. Yet, it only provides a perturbative
approach. Another drawback is that this approach only works for a coupling constant 1.
Figure 3.1: The standard model of particle physics. The above two rows are the quarks. These particles
are charged with respect to all three forces. The lower two rows are the leptons. The third row, the
neutrinos, carry only an isospin charge. The lower row, the charged leptons, also carry an electric charge.
The right column lists the gauge bosons. The photon is an excitation of the electromagnetic field, the Z 0
and W ± are the carriers of the weak interaction. The eight gluons carry the strong force (image from
Fermilabs, 1995)
3.2.1
Instantons
Using perturbation theory, it is not possible to write down interactions that are non-perturbatively in
nature. Such interactions that do not have an infinitesimal form. An example of such interactions is
a change of topology of the field. Solutions to such interactions are solitons, particle-like solutions
linking ground states of different topology. As an example, consider a map of U(1) to U(1). This map
has a non-trivial topology
eiφ → eniφ
(3.13)
with n the winding number. This number arises because the periodicity allows non-trivial equivalence
maps.
26
3.3 Confinement and asymptotic freedom
The instanton interaction arises because there does not exist a trivial map between the QCD symmetry group SU(3) and the S(3) symmetry of Wick-rotated spacetime at infinity (equation 3.15). As
this map is not only fixed in space but also in time, it represents a soliton solution fixed in time, hence
the name instanton. This instanton looks very much like an interaction. This is in contrast with the
topological solution resulting from equation 3.13. Such a topological solution gives rise to soliton
solutions, particle like solutions that carry the topological charge needed to transform any state to the
other.
It is possible to write down this topological charge in terms of the fields. The equivalence for the
winding number is the Pontryagin index qT (Kapusta, 1989, ch. 8.4)
Z
1
qT =
d4 x Tr[µνρσ F ρσ F µν ]
(3.14)
16π2
This is a gauge invariant quantity, it is therefore not possible to get rid of this Pontryagin index by
a gauge transformation. Using this equation it is possible to derive an interaction strength of this
instanton interaction for particles carrying a colour charge. It can be shown that the strength inversely
depends on the mass of the light quarks. Note that this topological charge is defined in Euclidean
time, related to Minkowski time by a Wick rotation
te = it
(3.15)
The effects of QCD can only be accurately calculated at high energies, where perturbative QCD
is applicable. In this domain the instanton effects are always dominated by perturbative corrections
(Kapusta, 1989, ch. 8.4) and are therefore ignored in this thesis.
3.3 Confinement and asymptotic freedom
The fact that the weak and strong interactions are not Abelian has important consequences. In electromagnetism, vacuum polarisation shields the charged particles, leading to a smaller coupling constant
for larger distances or lower energies. The energy dependence of the coupling constant is known as
the running of the coupling constant. However, due to the self-interaction of non-Abelian theories,
the coupling constant of non-Abelian theories behaves differently: it increases for larger distances!
For the weak interaction the mass of the weak gauge bosons limits this effect, as a mass of the gauge
boson gives rise to an exponential reduction of interaction strength at large distances. However, the
strong interaction is exchanged by massless gauge bosons (gluons), so the inverse relation between
coupling strength and energy is fully present in QCD.
Theories having this effect are called ‘asymptotically free’, as in the limit of infinite energy the
interaction strength goes to zero. However, as the interaction energy drops, the coupling constant
grows to high values. This suggests the existence of ”confinement”, the fact that at normal energies
quarks cannot exist as individual particles. There is convincing evidence that in the limit of low
energy, the field strength is independent of separation distance between the quarks. This leads to a
spring-like force, where energy and separation distance depend linearly on each other. Trying to break
a qq-pair by pulling them apart will increase the energy stored in the field, up to the moment where it
is energetically favourable to create a new qq-pair form the vacuum, leaving two qq-pairs.
For this reason, at low energies quarks cannot exist as free particles. This is why in everyday life
we see the world around us being build from mesons and baryons, particles made from respectively
two or three quarks. These particles have no net colour charge (they are ”white”). Mesons are thus
build up from a quark and antiquark from the same colour-anticolour, while baryons must consist
27
Chapter 3: Quantum chromodynamics at high densities
of three quarks with each a different colour charge (red, green or blue). The residual force of these
quarks provides for the nuclear binding between hadrons, just as residual electromagnetic forces in
molecular binding explains the Van der Waals binding. The Van der Waals-force results from the
spatial distribution of electric charges in atoms and molecules. Close to the atom this can provide
weak minima of the potential, allowing molecular binding at low temperatures.
The baryon binding is best described by pion exchange. Pions are massive mesons, carrying the
quantum numbers of a quark-antiquark pair. Forces mediated by a massive particle fall off exponentially, as described by the Yukawa-potential
V(r) = −g2
e−mr
r
(3.16)
Using this potential, it is in principle possible to write down an equation of state for a baryonic compact star. However, especially in dense baryonic matter there are many uncertainties as a consequence
of the renormalisation procedure, as well as the need to include many terms as V(r) becomes very
large for small r. As a result there are many different models for the baryonic potential (see Sedrakian
and Clark 2006 for an overview).
However, there exist situations where energies are not so low. One example is the early universe,
where temperatures and energies increase as the size of the universe decreases. This high temperaturelow baryon density limit can also be studied in particle accelerators, where the phase transition from
nucleons to the quark-gluon plasma has possibly been observed. Temperatures where deconfinement
occurs are of the order of 1012 K.
Another limiting case is the case of low temperature-high density. As densities grow, the separation distance between particles decreases and due to the exclusion principle the average particle
energy increases. This leads to the deconfinement of the nucleons and the possibility of quark matter.
The only possible objects with densities high enough to make such matter possible are compact stars.
It might be energetically favourable to form strange quarks at increasing chemical potential, but before before confinement is lifted. If this is the case, the presence of baryons containing strange quarks,
hyperons, can influence the equation of state.
3.3.1
Quantum chromodynamics at low energies
Since QCD is a strongly interacting theory at low energies, perturbative methods are inappropriate
for these energies. Therefore, other methods are required to do calculations at these energies. One
of the options is to tackle the problem of QCD in a non-perturbative way. This is usually done by a
method known as lattice QCD. Here, spacetime is considered to be a discrete medium. This approach
transforms integrals over all spacetime to discrete sums, which can be handled by computers. Of
course the discretisation of space-time destroys translation symmetries. This can be solved by, after
performing the integrals in the discrete algorithm, going back to a continuous spacetime again.
However, it is very hard to do lattice calculations at finite density. The chemical potential will
insert imaginary exponents leading to fluctuations that are currently not under control. Therefore, this
algorithm has up to now not resulted in usable calculations at high density. According to Ivanov et al.
(2005) it is possible to do lattice calculations based on QCD inspired models in which this problem is
under control. According to the authors the results rule out the existence of hybrid stars.
Another approach to tackle strongly interacting QCD is by the MIT bag model. Originally invented as a phenomenological model for protons, this model is also used for quark stars. The model
is based on the idea that in strongly coupled QCD the vacuum will be filled virtual quark-antiquark
pairs. These virtual pairs will produce a pressure, the bag pressure, confining the quarks in a cavity,
28
3.3 Confinement and asymptotic freedom
the bag. The pressure can in principle be calculated from QCD, but is mostly fitted to a desired size of
the bag. Generally accepted values for the bag pressure are around 150 MeV (DeGrand et al., 1975).
This bag approach is also used for quark stars. This clearly demonstrates that a quark star is merely
a huge baryonic particle, just as a neutron star is comparable to a huge nucleus. The disadvantage of
this approach is that it is highly dependent on the chosen bag pressure, which cannot be calculated at
finite density and energies below the scale where perturbative QCD becomes valid. At energies where
perturbative QCD does become valid, it is probably preferable to use that technique, as it is QCD.
A third approach is the NJL model by Nambu and Jona-Lasinio (1961). This model describes
the quark interaction as a four-fermion interaction. The gluon mediating this interaction is then contracted to a point. All the properties of the gluon interaction must then be described by the coupling
constant of the NJL interaction. The advantage of this theory is that it provides simplifications allowing calculations at lower energies than perturbative QCD. The disadvantages are twofold. First, it is a
non-renormalisable theory, allowing calculations only to a momentum cutoff. It can therefore not be
used to probe the high energy limit of QCD. It is thought that quark matter in quark stars lies below
this limit (Buballa, 2005). A second drawback is that it does depend on the phenomenological interaction strength of four quarks, something that is determined from the microscopic interaction including
gluons that actually takes place. This theory only provides a phenomenological description of QCD,
that still has to be fitted to the actual theory. It is very usable as a mechanism to calculate effects such
as colour superconductivity (see section 6.2).
3.3.2
Phase transitions in dense matter
Strangeness (ns/nu)
Going from nuclear matter to quark matter, a
phase transition has to occur where chiral sym1
metry is restored and confinement ceases to exist. Witten (1984) proposed the possibility that
0.8
this actually might be the lowest energy state
0.6
of matter at all densities. There is a potential
barrier blocking the conversion of baryonic mat0.4
ter to quark matter. If matter becomes dense
enough to overcome this barrier, the exothermic
0.2
energy of the reaction then ensures the conversion of all matter to quark matter. Depending
0
0
50 100 150 200 250 300 350 400 450 500
on the actual state of the matter, crusts of norµ (MeV)
mal matter might still be possible (Alford et al.,
2006b, Jaikumar et al., 2006b, Stejner and MadFigure 3.2: Strangeness as a function of chemical
sen, 2005), while the effects of pure quark matter
potential. Assumed is a mass difference between a
on the vacuum might seriously alter the specneutron and a hyperon of 200 MeV. Right of the
trum of objects made of quark matter (Usov,
vertical line deconfinement is assumed, and again
2001).
a mass of the strange quark of 200 MeV is assumed. This diagram is calculated assuming nonLess extreme is the idea that there exists a reinteracting particles, and a non-running strange
gion of supernuclear density where quark matter
mass. It is therefore not very accurate, and only
is more stable than nuclear matter. It is then posfor illustration.
sible to convert only part of the compact object
into quark matter. This leads to the possibility of
so-called hybrid stars, consisting of a quark matter inner core, a possible superfluid neutron matter
outer core, and all the rest of the layers expected in neutron stars. These stars are hard to distinguish
29
Chapter 3: Quantum chromodynamics at high densities
from ordinary neutron stars. However, there are some options. The presence of quark matter inside alters the mass-radius relation, and as the equation of state differs properties may vary. Expect different
cooling curves and a different behaviour of the magnetic field.
Another possibility is the formation of strange quarks before deconfinement occurs. The mass of
the lightest hyperon, the Λ-particle consisting of of uds quarks, is about 1150 MeV, 200 MeV more
than the mass of protons and neutrons. For chemical potentials > 200 MeV the exclusion principle favours the formation of hyperons. At this chemical potential quarks are still confined but the
strangeness of the matter already increases. In figure 3.2 a plot of the situation is made.
The nature of the phase transition is still an open question. It depends on the coupling strength
of the strong force, which is still poorly known at high densities. Around the transition, the strong
coupling constant will be of order 1, the order where perturbation theory will fail to provide answers.
There exist phenomenological theories, designed to describe features of QCD, which can be used to
study this regime. However, as these theories are not QCD, the validity of the results is uncertain.
Another approach is to consider spacetime as a discrete medium, a lattice. This turns the path
integrals over spacetime into sums which can be calculated. The advantage of lattice QCD, as this
approach is known as, is that it automatically includes momentum cut-offs providing regularisation.
Lattice QCD has proven useful in low energy calculations of QCD as well as high temperature calculations, but suffers from problems with high density calculations. These difficulties are known as the
fermion sign problem, the result of the partition function acquiring an imaginary part.
3.4 The running coupling constant in QCD
k0
k
k
k0
Figure 3.3: Loop corrections to the gauge boson propagator are one of the contributions that lead to an
energy dependent coupling constant
At high energies, the coupling constant decreases to values low enough for a truncated series
expansion to become a good approximation. The series expansion can be determined from the action
using the path-integral method. In (Freedman and McLerran (1977a) the propagators and vertices
are calculated, in Freedman and McLerran (1977c) the thermodynamic potential is calculated. In
these articles, the MOM renormalisation is used. To make gauge invariance manifest it is useful to
rewrite this to the MS scheme. Another reason to use the MS is that most QCD-parameters have been
calculated in the scheme.
One of the main problems of perturbative QCD is the renormalisation procedure. In QED, the
renormalisation of the boson propagator is defined as follows:
Π(q2 )renormalised = Π(q2 ) − Π(0)
(3.17)
Here Π(q2 ) is the full propagator. Defining it in this way, renormalisation is uniquely defined by
30
3.4 The running coupling constant in QCD
the requirement that electromagnetic cross sections to drop to zero at zero momentum transfer. Via
similar methods renormalisation uniquely defines the charge and mass in electrodynamics.
In contrast to electrodynamics, in QCD Π(0) is not defined. Therefore we are unable to uniquely
define the renormalisation procedure. The best we can do is to scale the running of the coupling
constant to an experimentally defined amplitude. However, the exact procedure is not well defined,
resulting in different outcomes for calculations in different schemes. Only when all terms in the series
expansion are included the scheme dependence will drop out.
3.4.1
The running of the coupling constant
A very powerful way to study the running of the coupling constant α(Λ) is via the renormalisation
group. The renormalisation group equation
M
∂g
= β(g)
∂M
(3.18)
g3
(11N − 2N f )
48π2
(3.19)
where β is, to lowest order,
β0 (g) = −
expresses the evolution of the coupling constant in a differential equation. As can be seen from this
equation, asymptotic freedom as discussed in section 3.3 occurs as long as β is negative. This is the
case when N f < 5.5N. Here N is the number of colours and N f the number of flavours. The equation
has the solution
g2
6π
α(Λ) =
=
(3.20)
4π (11N − 2N f ) log(Λ/Λ0 )
Using experimental data, Λ0 can be determined.
Yet, an uncertainty remains as α is specified as a function of Λ, the energy of in the gluon propagator carrying the energy q2 = Λ2 . However, this is not a useful quantity, as we are unable to determine
this energy. The relevant energy scale in our case will be the chemical potential µ. This can be solved
by replacing Λ → cµ, with c a number of order unity. This introduces a new scale dependence into the
equations. This parameter is the result of truncating the perturbation series expansion, and the scale
dependence can be understood as a result of not describing the microscopic physics accurately. More
on this in Kapusta (1989, ch. 4).
3.4.2
The running of the quantum chromodynamics coupling constant
Using the methods described in the previous section we can derive an equation for the coupling constant. Using not only β0 , but also β1 and β2 we can determine α s to three loop order (Particle Data
Group, 2004)



!2
4β21 
2β1 log(u)
4π 
1
β2 β0 5 

(3.21)
αs =
+ 4 2  log(u) −
+
− 
1 − 2
β0 u 
2
4
β0 u
β0 u
8β21
In this equation β0 = 11 − 2N f /3, β1 = 51 − 19N f /3 and β2 = 2857 − 5033N f /9 + 325N 2f /27, with N f
the number of flavours. Furthermore, u = log(Λ̄/ΛMS ). ΛMS can be determined using experimental
data. Requiring α s = 0.3089 at Λ̄ = 2 GeV, ΛMS = 365 MeVfor N f = 3 (Particle Data Group, 2004),
in the MS scheme, as a determination of Λ is approximation scheme dependent.
31
Chapter 3: Quantum chromodynamics at high densities
1
0.8
αs/π
Having defined α s this way, the only uncertainty left is the dependence of Λ̄ on µ. We take
the dependence of Λ on µ to be Λ(µ) = cµ, as
described in the previous section. The running
of the coupling is plotted in figure 3.4.
0.6
0.4
0.2
0
0
500
1000
Λ (MeV)
1500
2000
Figure 3.4: Running coupling constant to three
loop order (equation 3.21)
32
CHAPTER 4
Thermodynamics of degenerate quark
matter
A typical star consists of about 1060 particles. For so many particles, a statistical approach is the
only way to describe the system. Therefore, a thermodynamic approach to QCD is necessary. In this
chapter we will first look at the general properties of the Fermi-Dirac distribution describing noninteracting Fermi-particles. However, we will later see that the interaction plays a crucial role in the
realistic description of Fermi-matter in a star.
4.1 Fermi-Dirac distribution
One key property of fermions is that their wave function has to be fully antisymmetric. No fermion
may have the same set of quantum numbers as any other fermion in the system. This property is
known as Pauli’s exclusion principle.
Using this principle, one can write down a density of states for a non-interacting Fermi gas. We
assume a harmonic oscillator potential well, so all states are separated by an energy of ~ω. Each
individual state can only hold one fermion. The occupation number of such a state is thus 1 or 0.
This also implies that in the ground state the fermion with the highest energy must have an energy
of (n + 1/2)~ω, n ∈ N. In case of high temperature, when the average fermion energy is much
higher than that, the exclusion principle is not a severe limitation. In such cases there are much more
states available than there are fermions, and the distribution will approach the Maxwell-Boltzmann
distribution. However, in dense materials at low temperatures this limitation is very important. In
everyday materials such as iron at room temperature the electrons in the iron have average energies
equivalent to temperatures of order 5 × 105 K. In these cases the electrons will occupy all the low
energy states, with only a few empty states near the Fermi surface at (n + 1/2)~ω. The partition
function for one fermion energy level is
Zfree (i ) =
1
X
e−βns (i −µ) = 1 + e−β(i −µ)
(4.1)
n s =0
From this the Fermi Dirac distribution can be obtained. It is
n(i ) =
1
exp[β(i − µ)] − 1
(4.2)
33
Chapter 4: Thermodynamics of degenerate quark matter
In these equations i is the energy of a level, µ is the chemical potential and β the inverse temperature.
In the second equation the energy is used to label the different states. Note that as T → 0 this relation
approaches the step function
n(i ) = Θ(µ − i )
with Θ(x) is 1 for x > 0 and zero otherwise. From the partition function thermodynamic quantities
can be derived. When considering a macroscopic object, the size of the object is much larger than
all other physical length scales. Therefore, we can convert the sums over all states to integrals. It is
convenient to label these states by their Fermi momentum. Using this one obtains
Z 3
d p
i
=
(4.3)
3
2π exp[β(i − µ)] − 1
Z 3
1
d p
n=
(4.4)
3
2π exp[β(i − µ)] − 1
"Z 3
#
d p
1
P = T log
(4.5)
2π3 exp[β(i − µ)] − 1
Mass (MeV)
These integrals can be converted to integrals over when the density of states is known. For an ideal
relativistic gas the density of states is given by 2 = (p2 + m2 ).
At low temperatures, there is a sharp transition at = µ from almost all states being occupied to
all states being unoccupied. For a three-dimensional system there exists a three-dimensional volume
in phase-space in which all the occupied states of the system lie. The volume is bounded by the Fermi
surface. In case of a free Fermi gas this phase space volume will be one eighth of a ball, in more
general models the shape of the volume can be altered. However, the existence of a sharp transition
surface is a general feature of all Fermi systems at low temperatures.
For compact objects this is a fundamental
concept. When a star runs out of nuclear fuel,
400
ms
there is no energy generating mechanism avail350
able to provide a thermal gradient and thus a
300
pressure gradient to counter gravity. As such a
250
star radiates away energy, it has to respond by
200
contracting. This process continues until there is
150
a new pressure generating mechanism available
to sustain gravity. In white dwarfs this is the de100
generate pressure of electrons, while in neutron
50
stars degenerate neutrons and in quark stars de0
600 800 1000 1200 1400 1600 1800 2000
generate quarks deliver the pressure.
Λ (MeV)
As mentioned in section 2.1, degeneracy can
lift the Fermi level to relativistic energies. In
Figure 4.1: Running mass of the strange quark,
case of this relativistic degeneracy, the solution
calculated to one-loop order in alpha (the first two
of the structure equations becomes insensitive to
terms of equation 3.21)
central pressure and radius, indicating a maximum mass, the Chandrasekhar mass (see equation 2.23). Relativistic degenerate means that the mean kinetic energy per fermion is more than the
rest mass of that fermion. For a white dwarf to become relativistic degenerate the average fermion
energy thus has to be larger than 0.511 MeV.
In quark star models the chemical potential at the centre is of order 500 MeV, while up- and downquark masses are of order 5 MeV. Although the strange quark is much heavier, its energy dependent
34
4.2 The MIT bag model
mass is still well below this limit, for an assumed dependence Λ(µ) = cµ, c ∼ 2 (see section 3.4 for a
discussion). Therefore, quark star matter can be approximated by an equation of state for relativistic
degenerate fermions. In order to understand why solutions to the structure equations using a quark
equation of state are dependent on central pressure, we need to include interactions. To do this we
will first use a phenomenological model, known as the MIT bag model, to do some qualitative study.
Later we will derive interaction models based on perturbative QCD.
4.2 The MIT bag model
In the MIT bag model quarks are considered to be free particles confined to a bounded region by a bag
pressure B. This extra pressure term is a phenomenological description of the quark-quark interaction
binding the quarks. A convenient way to describe the thermodynamics of a system is to use the grand
thermodynamic potential Ω = −U − µn − T s, with U internal energy, n particle density and s entropy
density. Furthermore, µ is the chemical potential and T the temperature. From this potential quantities
as pressure, number density and energy density can easily be derived. They are
P(µ) = −Ω(µ)
∂p
n(µ) =
∂µ
(µ) = −p(µ) + µn(µ)
(4.6)
(4.7)
(4.8)
For the MIT bag model the thermodynamic potential becomes
Ω(µ) = −
N f µ4
1
+ B = − ( − 4B)
2
3
4π
(4.9)
The first part is the thermodynamic potential of a free Fermi gas, the second term is the bag pressure.
Accepted values of this bag pressure are of order 150 MeV (DeGrand et al., 1975).
As can be seen, the effect of the bag constant is to generate a vacuum pressure, making it possible
for the applied pressure to become zero before the energy density drops to zero. This in contrary to a
free gas where (P = 0) = 0. Using the bag model as an equation of state in the structure equations
makes it possible to find solutions which depend on central pressure and are thus mass dependent.
This is discussed in more detail in section 5.1.1.
4.3 Finite temperature field theory
In the previous chapter zero-temperature, zero density QCD was discussed. However, we need a
description of QCD applicable in dense matter. This requires a statistical approach. We will again use
the partition function
Z
Z = Tr exp −β(H − µi Ni ) =
Dφ hφ| e−β(H−µi Ni ) |φi
(4.10)
where D denotes a functional integration (see appendix B). In this equation β = (T )−1 and H is the
Hamiltonian of the system, as usual. By a Laplace transform this has the particle number operator
Ni as a conserved charge with the chemical potential µ as the Lagrange multiplier. The role of β is
similar: it is the Lagrange multiplier setting the mean energy of the system, with H as the conserved
energy.
35
Chapter 4: Thermodynamics of degenerate quark matter
From the partition function various thermodynamic quantities can be derived
∂ log Z
∂V
∂ log Z
Ni = T
∂µi
∂T log Z
S =
∂T
P=T
E = −PV + T S + µi Ni
(4.11)
(4.12)
(4.13)
(4.14)
The partition function, equation 4.10, can be used to write down the partition function for an ensemble
of particles. In this example I will discuss bosons. The expectation value for exp[−iHt] is
" Z t Z
!#
Z
Z
∂φ(x, t)
−iHt
3
hφ| e
|φi =
Dπ Dφ exp i
dt d x π(x, t)
− H(π(x, t), φ(x, t))
(4.15)
∂t
φ
0
which can be derived using the path integral method (see Kapusta, 1989). The trick is to use the
similarity between the time translation operator exp(−iHt) and the partition function Tr[exp(−βH)].
First we replace the time coordinate with an imaginary one: τ = it. Doing this in equation 4.15, it is
clear that β has the effect of changing the integration range of t. When T ↓ 0 ⇒ β → ∞, so in that
case the integration range for τ is [0, ∞). For T , 0 the integration range for τ changes to [0, β], whilst
requiring that φ(x, 0) = φ(x, β), a consequence of taking the trace.
Using all this, the partition function for bosons becomes
"Z β Z
!#
Z
Z
∂φ
3
Z=
Dπ
+ H(π, φ) + µN(π, φ)
Dφ exp
dτ d x iπ
(4.16)
∂τ
periodic
0
A similar technique can be applied for a fermion field, arriving at
! #
"Z β
Z
Z
0
3
0 ∂
Z=
DψDψ exp
+ iγ · ∇ − m + µγ ψ
idτ d x ψ −γ
∂τ
0
(4.17)
From these equations, one can evaluate the partition function and then determine the thermodynamic potential using perturbation theory. In section 4.3.2 we will see how to do this in the limit
where T = 0, µ , 0. First, let us look at some of the formalism required.
The partition function for an N particle system is
Z
Z = N DφeS (φ)
(4.18)
where the action S (φ) can be written as S (φ) = S 0 (φ) + S I (φ), with S 0 (φ) the free part quadratic in
the fields and S I (φ) the interactions of higher order. Using this, equation 4.18 can be written as
Z
∞
X
1 l
S 0 (φ)
Z = N Dφe
S (φ)
(4.19)
l! I
l=0
where S lI (φ) is the perturbative term of order l in the coupling constant. We can then take the logarithm
of this equation
R


!
Z
∞
S (φ) l
X

1 Dφe 0 S I (φ) 
S 0 (φ)


R
log Z = log N Dφe
+ log 1 +
(4.20)
l!
DφeS 0 (φ) 
l=1
36
4.3 Finite temperature field theory
Note that this equation is now split in a free and an interacting part
log Z = log Z0 + log ZI
(4.21)
The free part is the ideal gas contribution, as discussed in section 4.1. The interactions are written as
a perturbation series expansion. The actual calculation for a term of order l in the coupling constant
is then
R
D
E
DφeS 0 (φ) S lI (φ)
S lI (φ) ≡ R
(4.22)
0
DφeS 0 (φ)
For a coupling constant α/π < 1 the high order terms will be small, ideally going to zero as l → ∞.
This suggests that a good approximation can be obtained using only the terms S lI (φ) with l < lterm ,
where lterm is the limit of the expansion chosen such that the desired accuracy is achieved.
However, the number of possible interactions in S lI (φ) increases with l. For a series expansion
of S lI (φ) the best that can be achieved is that the series expansion converges, the sum of the asymptotic series expansion generally does not. The best accuracy that can then be achieved is the series
expansion up to lterm ≤ lmax , where lmax is determined by the condition
l=l
max D
X
S lI (φ)
l=1
E
0
<
l=lX
max +1 D
S lI (φ)
E
0
(4.23)
l=1
As an extra limit to this is the computational effort required for higher order terms. These terms
generally contain contributions of many different possible interactions, resulting in a limit that is in
practice a far stronger limit than the limit in equation 4.23.
4.3.1
The thermodynamic potential
In thermodynamics, the thermodynamic potentials are the potentials of the different ensembles. In
this thesis the ensemble used is the grand ensemble. Therefore the related thermodynamic potential,
the grand potential, will be referred to as ‘the thermodynamic potential’.
Ω = U + T s + µi Ni = β−1 log Z
(4.24)
Since Ω ∝ log Z, the above equation can be handled perturbatively in the region of small coupling
constant, as shown in equations 4.18-4.22. Since log Z is now described in terms of the action, standard field theoretical methods can be used. However, in a non-zero temperature or density, extra terms
appear. This is the result of infrared divergences, which give zero at zero temperature or density, but
yield a finite value if the temperature or chemical potential cannot taken to be zero (Kapusta, 1989,
ch. 3)
When T = 0 but µ , 0 terms of order q4 log q2 will appear (Kapusta, 1989, p. 75). These terms are
the effect of corrections to the gluon self energy interacting with a thermal or dense vacuum, resulting
in an effective mass for the gluon. As the perturbation is done assuming zero gluon mass, extra
perturbative terms compensating this arise. The dense interactions can be written as ring diagrams or
plasmon corrections (Gell-Mann and Brueckner, 1957). These diagrams can be summed up to obtain
a term that is not an integer power of α. This term has to be added to the series expansion obtained
from equation 4.22.
This effect is present in QED, where the plasmon contribution is the same as in QCD. We will do
the calculation for QED, which is much easier but leads to the same result (Kapusta, 1989, ch. 8).
37
Chapter 4: Thermodynamics of degenerate quark matter
Π
− 12 α2s
Π +...
Π + 13 α3s
Π
Π
Figure 4.2: Ring diagrams, where α s is the strong coupling constant.
The contribution from the ring diagram can be written as
Z
log Zring
1 X
d3 k
Tr log[1 + Π0 (k)Π(k)] − Π0 (k)Π(k)
=− T
3
βV
2 n
(2π)
(4.25)
with the photon propagator Π0 given by
µν
Π0 =
and Π(k2 ) given by
1
1
ρ kµ kν
µν
µν
P
+
P
+
G(kµ ) − k2 T
F(kµ ) − k2 L
k2 k2
−1
Πµν = Π−1
µν − Π0µν
(4.26)
(4.27)
0
with Π−1
µν the inverse of the full propagator. Furthermore, F and G are scalar functions of k and |k|,
PT and PL are transversal and longitudinal projection operators and ρ is a gauge fixing parameter
kµ kν Πµν = ρ. Using these equations equation 4.25 can be rewritten to
"
#
"
#
!
Z
log Zring
1 X
d3 k
G(n, ω)
G(n, ω)
F(n, ω)
F(n, ω)
=− T
2
log
1
−
+
2
+
log
1
−
+
βV
2 n
(2π)3
k2
k2
k2
k2
(4.28)
F and G are functions of n and ω as k0 = 2iπnµ and ω = |k|. This function is not finite, and to isolate
the infinities you can calculate equation 4.28 at ω = n = 0. The finite term is then the difference
between equation 4.28 at ω = n = 0 and ω, n , 0. The resulting equation yields

!2
!2 
Z

log Zring
F(n, ω) 
1 X
d3 k 
 G(n, ω)
+
=− T
2



βV
4 n,0
(2π)3 
k2
k2
!2
!2
!2
!2 

G(0, ω)
G(0, 0)
F(0, ω)
F(0, 0) 

+2
−2
+
−
(4.29)


2
2
2
2

ω
ω
k
ω
As T = 0 in our case, it is useful to perform a Wick rotation k4 = −ik0 and to introduce k2E =
k42 + k2 = −k2 ≥ 0. Note that β ∝ T −1 , so equation 4.29 does not yield 0. Since F and G are functions
of kµ , these functions now depend on the Euclidean coordinates. Independent coordinates are |kE | and
φ with tan φ = |k|/k4 . Using all this we can transform the ring contribution to



Z 1/2π
Z
G(k2E , φ) 
G(k2E , φ)

log Zring
1 3 ∞ 2 2
2 
 − 2
=−
dkE kE
dφ sin φ 2 log 1 +
βV
(2π) 0
k2E
k2E
0



F(k2E , φ)  F(k2E , φ) 




+ log 1 +
(4.30)
−
k2E
k2E
38
4.3 Finite temperature field theory
When using the explicit forms of F and G (see Freedman and McLerran, 1977b). one can obtain a
ring contribution in the relativistic limit
log Zring
e4 log e2 4
=−
µ
βV
128π6
(4.31)
As noted, in QCD this result is still valid and give rise to terms of order g4 log g2 (see Freedman and
McLerran, 1977c).
4.3.2
Thermodynamic quantities of a quark gas
k0
k0
k
α2
k
+α2
k0
k
k0
k
k
k0
+α2
k0
k
k
k0
+perm.
+α2
k0
k
Figure 4.3: Two particle interaction to second order with renormalised propagators
k00
−α2 k0
k
k
k00
k
2
k00 −α k0
k00 +perms.
k0
k0
k
Figure 4.4: Three particle interactions with renormalised propagators
Using the methods described above it is possible to derive the thermodynamic potential for a gas
of free quarks. This is done in Freedman and McLerran (1977c) and Baluni (1978) in the MOM
subtraction scheme. The terms included are the ring diagrams as given above, the two-particle to twoparticle transition diagrams to second order in the strong coupling α s and three-particle interactions.
In figures 4.3-4.4 a schematic overview of the contributing interaction diagrams is provided. Not
included in these figures are all permutations of the diagrams. Furthermore, we need to take the trace
of the diagrams and integrate over all momenta. This as equation 4.22 expresses vacuum to vacuum
expectation values, while the diagrams represent particle interactions.
The total thermodynamic potential is given by
Ω = Ω(0) + Ω(1) + Ω(2) + O(α3s )
(4.32)
with Ω(n) the thermodynamic potential correction of order αns . The zero and one loop correction
39
Chapter 4: Thermodynamics of degenerate quark matter
potential for both massless and massive quarks for T = 0, µ , 0 is (Baluni, 1978)
!
m2a µa 4
4 2
Ω = Df
(− π )φ0 2
3
µa 2π
a
!
X
m2a µa 4
(1)
Ω = D f C f αs
(2π)φ2 2
µa 2π
a
X
(0)
(4.33)
(4.34)
and φ0 and φ2 are given by
"
#
5
3 2
1 + (1 − x)1/2
φ0 (x) = (1 − x) (1 − x) + x log
2
2
x1/2
(
"
#)
1 + (1 − x)1/2
φ2 (x) = 3 (1 − x1/2 − x log
− 2(1 − x)2
1/2
x
1/2
(4.35)
(4.36)
The second order term has only been calculated for a massless quark
!
!
!) 4
X (5
X
1
21
µa
2µ2 µa 4
2
(2)
(2)
2
− D f C f αs
Cc + C f − Cc π +
− 165
Ω = D f C f β αs
log
eM 2π
8
2
4
2π
a
a
(4.37)
with
(
)
11
4
(2)
β (α) = − Cc + N f D f C f /Dc α2
(4.38)
33
3
Since the second order contribution for massive quarks is unknown, we have to take Ω(2) = 0 for
massive quark flavours.
In the above equations Cc, f and Dc, f are the Casimir numbers for respectively the colour and
flavour symmetry group. Nc, f are, respectively, the number of colours and flavours. More specific
2NC f = Dc = N 2 − 1,
Cc = D f = N
(4.39)
In nature Nc = 3. The value of N f depends on which quark flavours contribute to the diagrams. As
the quarks all have different masses it can be seen as an energy dependent value. At low energies only
two quarks will be present, so N f = 2 in that case. In quark stars the chemical potential will be high
enough to overcome the strange quark mass, so in quark stars N f = 3, as forming strange quarks will
lower the Fermi levels and thus be a state of lower energy. The charm, bottom and top quarks are
much heavier than the up, down and strange so it is not energetically favourable to form these quarks.
The above formulae are in the momentum-space subtraction (MOM) scheme. Much more common is the MS scheme. This scheme has the advantage of preserving manifest gauge invariance. The
relation between MOM and MS is
αMOM
αMS s
= s 1 + AαMS
s
π
π
(4.40)
with A = 151/48 − (5/18)N f . This leads to a thermodynamic potential to second order for massless
quarks of
(
!
!
)
α α N f µ4
2
Λ̄ α s 2
s
s
Ω(µ) = − 2 1 − 2
− G + N f log
+ 11 − N f log
π
π
3
µ π
4π
40
(4.41)
4.3 Finite temperature field theory
with G = G0 − 0.536N f + N f log N f where G0 = 10.374 ± 0.13 1 . Λ̄ is the renormalisation subtraction
point.
It is also possible to include the nonzero strange quark mass. In that case the equation for the
massive quark becomes, to first order (Fraga and Romatschke, 2005)
!
"
2
µ + u # α (N 2 − 1) " µ + u
Nc
3 4
5 2
s c
2
2
+
− µu
Ωmass. = −
µu µ − m + m log
3 m log
2
2
m
m
12π2
16π3
!
#
µ + u Λ̄
−2µ4 + m2 6 log + 4 µu − m2 log
(4.42)
m
m
p
with u = µ2 − m2 and m the energy dependent running mass
m(µ) = m̂
α 4/9 α
s
1 + 0.895062
π
π
(4.43)
with m̂ ' 262 MeV. As m depends on α s , the scale dependence of α s is inherited by m. Note that, in
the limit of m ↓ 0, equation 4.42 reduces to the first two terms of equation 4.41 , as it should.
1
The uncertainty in G0 is from errors in numerical integration as performed by Freedman and McLerran (1977c)
41
CHAPTER 5
Mass-radius relation of quark stars
5.1 Numerical stellar models
As discussed in section 2.1, stellar models can be described using the following set of equations: an
equation describing the pressure gradient given a certain distribution of matter and a set of equations
describing what the pressure will be given a matter distribution. Together these equations can be
integrated and yield a model of a star.
The first equation, that of how the matter is distributed given a thermal gradient, is called the
equation of hydrostatic equilibrium. In this thesis we will use it in relativistic form, the TOV equation
(equation 2.1). The set of equations describing the state of matter given an energy distribution is
called the equation of state, an equation that relates pressure, energy density and temperature. For
nonradiating objects these equations are enough to describe the system. For radiating objects like real
life stars radiation pressure might be important in certain regions of the star, and must therefore also
be included.
In the case of neutron stars and quark stars, the only form of radiation is due to cooling, and is only
important for the stellar structure during and just after the collapse. During these moments, the star is
very far from equilibrium, not justifying the use of the equation of hydrostatic equilibrium in the first
place. These situations are therefore computationally very much more complex than the situation of a
settled star in equilibrium.
But even in equilibrium, only the very simplified scenario using Newtonian gravity discussed in
section 2.1.1 has analytic solutions. In general relativity it is not possible to analytically solve the set
of equations 2.1 and 2.2 with the polytropic equation of state (equation 2.6). Yet including general
relativity but leaving out other interactions is not good enough for neutron stars and quark stars. Dense
matter in neutron stars and quark stars is characterised by strong internal interactions, so an ideal gas
approach is insufficient. Instead of just guessing a good polytropic index γ it is therefore better to
explicitly account for these interactions.
When calculating models of quark stars temperature can generally be ignored, as the binding
energy is many orders of magnitude larger than the thermal energy. The remaining set of equations to
be solved contains only two differential equations and one equation linking energy density to pressure,
requiring two boundary values. Typical boundary values would be to require that r(P = 0) = R∗ , where
R∗ would be a user-supplied value. Another natural boundary would be to replace the radius with the
total mass of the star, still requiring the pressure to vanish at the surface. In the latter approach,
however, the radius is not a pre-determined condition but is calculated by the vanishing-pressure
43
Chapter 5: Mass-radius relation of quark stars
requirement.
These approaches work in principle, but are computationally very difficult. The most straightforward choice for boundary conditions is to start calculating from an initial central pressure and keep
iterating through the equations until the pressure is zero. Again r(P = 0) = R∗ and 4/3πR3∗ ρ̄ = M∗ ,
but both parameters have become internal variables. Calculating such models for a range of initial
pressures produces outcomes of models with different masses and radii. These outcomes can then be
combined in a mass-radius diagram.
Typically when the central pressure is increased, the mass will be increased. This can be understood as the central pressure is caused by the weight of all the mass on top of it. If mathematical
solutions give other results, these results are not physical.
It is possible that for a given central pressure there are multiple solutions for central energy. White
dwarf stars, for example, can have masses that are also possible for neutron stars. Unless of cosmological origin, nature will always favour the first stable solution, as all concentrations of mass have
formed as a result the collapse of gas in the interstellar medium, a very dilute material. As knowledge
of white dwarf structure is well-established, and the predictions on the maximum mass of these objects are all of the order of the Chandrasekhar limit, neutron stars or quark stars much lighter than the
Chandrasekhar limit will be considered as unphysical. Note that rotation provides an extra balancing
force against pressure and is thus not a mechanism of creating underweight neutron stars or quark
stars.
The maximum mass of an object can be found using a plot of the mass radius diagram. Most of
these diagrams will show a peak in mass at a certain radius, allowing solutions for all masses smaller
than Mmax . Increasing the central density further after Mmax leads again to smaller masses. This is
a signature of reducing mass while increasing the central pressure and is unphysical. All solutions
beyond R∗ (Mmax ) where the tangent vector to the mass-radius plot is negative should therefore be
seen as a mathematical rather than a physical solution.
5.1.1
The MIT bag model
As discussed in section 2.1.1 it is possible to write the interactions as an extra energy term contributing
to the internal energy. Using the MIT bag pressure (section 4.2) as the extra energy term, it is possible
to study the general behaviour of massless quark star models. The MIT bag model will provide a
vacuum pressure B that makes it possible that solutions to equations 2.1 and 2.2 span a range of
masses, contrary to the solution in equation 2.23, which uniquely defines a maximum mass.
The internal energy is changed by the presence of a bag pressure by
∆EMIT = −
4B
3n()
(5.1)
indicating that when P = 0, there is still an energy density of 4/3 B. This has the effect of binding
the star tighter together, just as a general relativistic correction does. However, the change in energy
∆EMIT is independent of the density and radius, in contrast with ∆EGR , which does depend on both
density and radius.
Combining the above equation with equations 2.24 and 2.25 the equation for energy balance
becomes
2
N 1/3 GN p
4B
2
7
E=
−
− 0.918294M /3 ρc/3 −
=0
(5.2)
r
r
3n( p )
Here the Newtonian gravitational potential is somewhat strange, as massless particles are assumed.
Using E = mc2 it is possible to use the particle energy as its mass. Furthermore, the equation for
44
5.1 Numerical stellar models
the general relativistic correction was derived using a non-interacting gas, so the term will differ from
the term in this equation, as discussed below equation 2.25. I have used the energy per particle as
parameter. The energy per particle is
3µ
Bπ2
p =
(5.3)
+ 3
4
µ Nf
Solving equation 5.2 for r yields
√3
r≈
N − GN p2
0.92ρc/3 M 7/3 +
2
4B
3n( p )
(5.4)
It is clear that both the general relativistic correction and the bag pressure change the situation of section 2.1.1, where no solution existed for a relativistic gas. The effect of general relativity is to decrease
the maximum stable mass somewhat. This can be understood in that general relativity increases the
potential, so a collapse will occur earlier. The bag pressure also allows a smaller maximum mass and
radius, for the same reason as with general relativity. So models with interaction generally lead to
solutions with a smaller radius than models without any interaction.
It is interesting to note that the bag pressure does not depend on any stellar parameter. For a very
light star, the bag pressure is the same as for a more massive star. For models of a light star, the bag
pressure is then the dominant contribution, so we can ignore the gravitational potential
√3
3 Nn( p )
r=
4B
(5.5)
This equation scales as r ∝ N 1/3 , the scaling law of the mass-radius equation for an incompressible
medium. The presence of the particle density is not important for small objects as the particle density
barely changes over the object.
For model calculations equation 5.2 for energy density instead of energy per particle is probably more useful. From this equation it is clear that in the low mass scenario the energy density is
independent of the radius, a signature of an incompressible medium.
4
GM(r)
2
7
E = free − B −
− 0.918294M /3 ρc/3 = 0
3
r
(5.6)
This equation uses again a non-relativistic gravity with a correction term for models using an ideal
gas equation of state.
Only at very high mass general relativity starts to play a role. It is quite easily seen that general
relativity steepens the gradient in the mass-radius plot. At high masses the numerator of equation 5.4
decreases due to the term ∝ N 2f = M f while the denominator increases as M increases. As expected,
both the higher energy density and the greater mass contribute to the steepening of the potential. The
general relativistic correction term assumes an ideal gas equation of state. Extra attractive interactions
result in smaller stars, increasing the effect of general relativity. This can only be calculated using a
general relativistic equation for gravity.
The change to a polytropic equation is that the pressure vanishes at finite energy density. In other
words, the polytropic equation (equation 2.6) for a relativistic degenerate gas changes to
1
P = ( − 4B)
3
(5.7)
45
Chapter 5: Mass-radius relation of quark stars
where is the energy density without the energy density from interactions. Also from this equation
it can be understood that the Lane-Emden equation (equation 2.19) has more than one solution for a
relativistic gas when a bag pressure is included. Equation 5.7 can be rewritten to
4
1
P− B= 3
3
(5.8)
indicating that the bag pressure can be seen as an alteration of the boundary condition. For an ideal
relativistic gas the boundary condition is P = 0, while for an interacting gas with a phenomenological
bag pressure as interaction term the boundary condition changes to P = 4/3 B. While for a relativistic
gas P ↓ 0 only as r → ∞, r remains finite when a nonzero boundary condition P can be used.
5.2 Numerically solving models of quark stars
5.2.1
The method
As the TOV equations combined with a quark matter equation of state does not have an analytical
solution, I have used a computer programme to numerically solve the set of equations. The program is
written in C, and uses the Bulirsch and Stoer method described in Numerical Recipes in C (Press et al.,
1996) to integrate the set of differential equations. This method uses a rational function extrapolation
to describe the equations and integrates these. When combined with an adaptive step size method
based on a required precision this model produced accurate results in a reasonable time span (see
figures 5.9 and 5.11).
The integration sequence starts with an initial pressure, or chemical potential which I used as
running variable. Using this chemical potential the pressure and energy density are calculated. These
results are fed in the TOV equation, resulting in a pressure gradient. Using the relation
dµ dP dµ
=
dr
dr dP
(5.9)
the pressure gradient can then be converted back into a gradient in chemical potential.
Using this gradient, a next step δx from the origin can be calculated. The size of δx is determined
by the slope of the gradient: the steeper the gradient the smaller the step size. This algorithm not just
allows for a good precision and relatively low computing power, it also guarantees that the precision
is constant throughout the model.
This process is repeated until P(µ) = 0. The mass and radius are then given by
X
R∗ =
δx
(5.10)
all steps
M∗ = 4πr2 m(r)
(5.11)
P
which are both easily obtained from the calculations: r = δx is the running parameter used to
iterate through the star, M is the solution of one of the equations of the model, the equation of mass
conservation, at r = R∗ .
When this process is repeated for a range of initial chemical potentials a mass-radius relation is
obtained. I have used as a lower limit for the central chemical potential µc = µ(p = 0), for which one
expects a star with zero radius. The higher limit I have set by hand, such that the the most massive
solution of the model is included.
46
5.2 Numerically solving models of quark stars
5.2.2
Equations of state used in the model
Figure 5.1: Results of compact star models using
an equation of state of quark matter matched to
hadronic matter. The mass-radius for quark star
models are the same as in section 5.4.1. In this
plot ρ0 ∼ 0.48 fm− 3 is quark number density at
nuclear density (image from Fraga, 2006).
Using the method described above, the physics
lies in the TOV equations and the equations of
state used. I have based the calculations on
the assumption that quark matter does exist and
makes up the whole star. I have not included the
possibility of a hybrid star, where a normal neutron exterior would be accompanied by a quark
matter core. Although it is rather straightforward
to calculate mass-radius relations of hybrid stars,
I don’t consider it very useful as both the quark
equation of state and the high density neutron
equation of state are poorly constrained. Models
of both quark stars and neutron stars are strongly
dependent on the equation of state. Combining
the two only introduces as an extra uncertainty
the pressure at which confinement occurs and
thus increases the degeneracy of the model. Further study to the equation of state of both quark
matter and neutron matter would be necessary to
improve this situation.
The quark equation of state used is based on
a perturbative calculation of the thermodynamic
potential of a quark gas at zero temperature. The
thermodynamic potential uniquely relates P and ρ, and is therefore very suitable as an equation of
state. In the previous chapters it is described how to calculate such a thermodynamic potential. The
thermodynamic potential is calculated to second order in α s for massless, and to first order in α s for
any number of massive quarks. I have performed model calculations up to second order, and also
studied the effect of one massive quark species, up to first order.
As discussed in section 3.3.1, it is also possible to study quark matter using a phenomenological
model. Two often used models are the MIT bag model and the NJL model. The MIT bag model
simulates confinement by the introduction of a vacuum pressure, the bag constant, that act as a force
on the quarks confining them to a region in space. This has the effect of shifting the line in the
pressure-density diagram such that at zero pressure there is still a finite energy density. As can be seen
in figure 5.4 this is not very far from the truth. However, the bag pressure is not determined by the
model and has to be fitted to observations. As there currently do not exist observations of very dense
quark matter, it is difficult to establish proper bounds on the bag pressure. The bounds currently used
are based mainly on stability criteria for normal nuclei and input from other methods.
The NJL-model is an effective field theory describing the quark interaction as a four-fermion
interaction. Since there is no way to derive the interaction strength of the four-point interaction from
the theory, and introduces another free parameter. Also for this model the free parameters can be fitted
to the hadron spectrum.
47
Chapter 5: Mass-radius relation of quark stars
5.2.3
The programme
The code of my programme can be found at http://www.science.uva.nl/˜srhardem/code.
html. The programmes work by calculating a number of models from a lower chemical potential
to an upper one. Between these bounds the programme calculates a logarithmic distribution as that
guarantees an even spread of points in the mass-radius relation.
For each model, the method described above is used to calculate the pressure and density, which
are an input for the TOV equation. This process is iterated until P = 0. For the model including a
massive quark species, somewhat more work had to be done. As one quark gets a mass, the quark
matter is no longer strictly neutral by itself. This has to be compensated by adding an electron-content.
Also, there are now two different chemical potentials. All in all, charge neutrality requires
2
1
1
f (µu , µd , µ s , µe ) = nu − nd − n s − ne = 0
3
3
3
(5.12)
where na are the number densities of the specified particles
na =
dΩ(µa )
dµa
(5.13)
The chemical potentials can be constrained by beta equilibrium. From the reactions ud → uu + e− + νe
and u s → uu + e− + νµ the chemical potentials must satisfy
µd = µ s = µu + µe + νe,µ
(5.14)
As neutrinos can easily escape the quark star, their chemical potential is put to zero.
If we write µd = µ s = µ, µu = xµ and µe = (1 − x)µ, we can plug this into equations 5.12 and 5.13
and solve the resulting equation f (µ, x) = 0 numerically. In the programme the solving algorithm used
is the Golden algorithm of Numerical Recipes in C (Press et al., 1996). In a certain range of chemical
potentials there are multiple values of x, where x is a parameter that multiplies µ to obtain µu (see
figure 5.2). Following x as the chemical potential decreases easily identifies the point in the diagram
that has to be used, but sometimes the solving algorithm fails in this task. Improving the algorithm
probably solves this problem, but simply allowing some errors and removing those was much quicker.
5.3 The equation of state from perturbative quantum chromodynamics
Before we move to solutions of the TOV equation, we first discuss the equation of state of quark
matter obtained using perturbative QCD. In this section, we will discuss the equation of state obtained
assuming three massless quarks up to second order, and compare it with an equation of state of first
order where one massive quark is assumed.
5.3.1
Equation of state of three massless quarks
In figure 5.4 are the pressure-density diagrams obtained from the equation of state assuming three
massless quarks to second order perturbation theory. In figure 5.3 are the energy density and pressure
of a quark gas relative to an ideal gas plotted against chemical potential. The behaviour of the coupling
constant depends on the variable Λ, while the only physical parameter of the system is the chemical
48
5.3 The equation of state from perturbative quantum chromodynamics
2.5
µ=375 MeV
2
f(µ,x)
1.5
3
f(µ,x)
2
1
0
1
0.5
-1
-20
0.25
x
0.5
0.75
300
350
450
400µ (MeV)
0
-0.5
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
x
(a)
1
(b)
Figure 5.2: Values of the number density equation 5.12, rewrited as described below equation 5.14, as
function of x and µ (left pane) and as function of x at a fixed value of µ = 375 MeV (right pane). The
solving algorithm should find the rightmost root.
1.4
1.2
1
1
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0
0
200
400
600
µ (MeV)
800
(a) Pressure as function of chemical potential
Λ = 2µ
Λ = 3µ
Ideal gas
1.2
ε/ε0
P/P0
1.4
Λ = 2µ
Λ = 3µ
Ideal gas
1000
0
0
200
400
600
µ (MeV)
800
1000
(b) Energy density as function of chemical potential
Figure 5.3: Pressure and energy density calculated from the thermodynamic potential obtained by perturbative QCD to two loop order (equation 4.41).
potential µ. As discussed in section 3.4, the best we can do is to assume a relation Λ = cµ, with c a
free parameter. This parameter can then be fitted by other methods.
It is clear that for a finite chemical potential, the pressure vanishes while the energy density remains finite. This behaviour can also be obtained by using a bag model. However, in a bag model
the absolute difference between and P is constant, while in our approach the difference P − may,
and indeed does, vary. As the variation is not very large, it can be concluded that the MIT bag model
is a reasonable model for quark matter. Another reassuring observation is that for infinite chemical
potential the equation of state approaches the equation of state of an ideal gas, as seen in figure 5.4.
This shows that asymptotic freedom indeed occurs, as can also be concluded from figure 3.4.
Finally, when we plot the interaction strength versus the chemical potential (see figure 3.4) we
can see that for Λ & 548 the pressure vanishes before α < 1, so the use of perturbative QCD to the
surface of the quark star models is allowed for both the model using Λ = 2µ and the model using
Λ = 3µ. These chemical potentials are used as their pressure-density relation compares to that of the
MIT bag model with values close to the accepted bag pressure B1/4 ' 150 MeV. In Fraga et al. (2001)
49
Chapter 5: Mass-radius relation of quark stars
1000
Ideal gas
Λ = 2µ
Λ = 3µ
900
800
P (GeV/fm3)
700
600
500
400
300
200
100
0
0
500
1000
1500
3
2000
2500
3000
ε (GeV/fm )
Figure 5.4: Pressure as function of energy density. Note that the curves with interaction can be quite
accurately described by an ideal gas minus a vacuum pressure. This indicates that the MIT bag model is
a reasonable approximation for quark matter. See equation 5.15 and below for a discussion.
it is shown that the models compare very well to a non-ideal bag model
Ω(µ) =
N f µ4
αeff + Beff
4π
(5.15)
/4
with Beff
= 199 MeV and α s = 0.628 for the model using Λ = 2µ. The model using Λ = 3µ compares
1/4
to Beff = 140 MeV and α s = 0.626.
Higher values of c in Λ = cµ represent weaker interactions. This follows from the lower effective
bag pressure needed to fit this model to equation 5.15, but can also bee seen in figure 5.4. In this
figure, the line representing Λ = 3µ lies much closer the the line representing ideal gas. Higher values
of c 3, the effective bag pressure becomes too low to explain the low energy behaviour of QCD.
For c 2 the pressure of quark matter remains small for even very high densities, which is also in
contrast with calculations.
1
5.3.2
Equation of state for one massive quark flavour
In figure 5.5 the pressure and energy density as a function of chemical potential are plotted for one
massive and two massless quark flavour to first order in α s . In figure 5.7 the pressure is plotted as a
function of energy density. From the plots we see that the strange quark mass has influence only for
low chemical potentials or low pressure and energy density. this is expected for two reasons. First, as
the chemical increases, so does average momentum of the particles. Mass terms can only be ignored
if the momentum k2 m2 , which is the case only for high chemical potential. The second reason for
the massless approximation to become better at high chemical potential is illustrated in figure 4.1. The
running mass of the strange quark decreases as a function of chemical potential. As at high chemical
potentials the strange quark mass is lower, ignoring this mass term for large chemical potentials is
better justified.
50
5.3 The equation of state from perturbative quantum chromodynamics
ms = 0, running α
Running ms α
Ideal gas
1
1
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0
0
ms = 0, running α
Running ms α
Ideal gas
1.2
ε/ε0
P/P0
1.2
100 200 300 400 500 600 700 800 900 1000
µ
(a) Pressure as function of chemical potential
0
0
200
400
600
µ (MeV)
800
1000
(b) Energy density as function of chemical potential
Figure 5.5: Pressure and energy density as function of chemical potential for an equation of state assuming
one massive and two massless quark flavours (equation 4.42) compared to the equation of state for three
massless quarks. Both equations of state are calculated for Λ = 2µ, from a thermodynamic potential with
corrections up to one loop order.
1.2
1
ε/ε0
0.8
0.6
0.4
0.2
0
0
200
400
600
µ (MeV)
800
1000
Figure 5.6: Energy density as a function of chemical potential, calculated by the analytical calculation
(equation 5.21).
51
Chapter 5: Mass-radius relation of quark stars
1000
Ideal gas
Running ms
ms = 0
P (GeV/fm3)
800
600
400
200
0
0
500
1000
1500
3
2000
2500
3000
ε (GeV/fm )
Figure 5.7: Pressure as function of energy density for an equation of state assuming one massive and 2
massless quark flavours, compared to the case of three massless quark flavours. Both equations of state
are computed for Λ = 2µ, from a thermodynamic potential with first order loop corrections.
These plots differ from the analysis in Fraga and Romatschke (2005), where I think there is an
error in the energy density plot. I have checked this using an analytical calculation of the massless case
to first order. I compared these with the results both of my calculations and the results in Fraga and
Romatschke (2005). The numerical calculations for three massless quark species to one loop order
have been executed with the same programme as the numerical calculations for one massive quark,
except that I put the scale factor m̂ in the running mass (equation 4.43) to a tiny number. I have also
checked that this agrees with the result of using the massless first order equation for all three quarks.
The energy density can be calculated analytically. The energy density is given by
(µ) = −P(µ) + µn(µ)
(5.16)
with
P(µ) = −Ω(µ)
∂Ω(µ)
n(µ) =
∂µ
(5.17)
(5.18)
The thermodynamic potential for a noninteracting gas is given by
Ωf =
µ4
4π2
(5.19)
and Ω for one flavour is given to one loop order by
!
!
µ4
α s (µ)
α s (µ)
Ω= 2 1−2
≡ Ωf 1 − 2
π
π
4π
52
(5.20)
5.4 Numerical solutions to quark star models using a perturbative QCD equation of state
2.5
Λ = 2µ
Λ = 3µ
Mass (M)
2
1.5
1
0.5
0
0
2
4
6
8
Radius (km)
10
12
14
Figure 5.8: Mass-radius plot obtained by solving the structure equations 2.1 and 2.2 and the equation of
state of three massless quarks (equation 4.41) to two-loop running.
Using this we can write an expression for / f
2
= 1 + (−3α s (µ) − µα0s (µ))
f
3π
(5.21)
Using α(Λ) to one loop order, with a scaling factor Λ = 2µ one obtains an energy density as plotted
in figure 5.6. This clearly coincides with the numerical result of figure 5.5(b).
5.4 Numerical solutions to quark star models using a perturbative
QCD equation of state
In this section I will present the results of numerical solutions of the TOV equations using an equation
of state based on perturbative QCD. As we will see, the model strongly depends on the renormalisation
scale Λ. Unfortunately, there are few successful lattice QCD calculations that can be used to fix Λ,
as lattice QCD at high density is plagued by large imaginary components of the chemical potential to
which there is currently not a good solution (see Ivanov et al., 2005, for a lattice calculation of a QCD
inspired model).
5.4.1
Quark star models using an equation of state of three massless quarks
Using the results of section 5.3.1 it is possible to solve the TOV equations (equations 2.1 and 2.2).
This can only be done numerically. Using the computer programme described in section 5.2.3 I have
obtained the results plotted in figure 5.8. This figure is plotted with a precision on µ and M of 10−4 .
The radius is a running variable so its precision cannot be specified. However, as shown in figure 5.9,
a comparison between the precision I used and a precision 10 times less shows that the error is very
small.
53
Chapter 5: Mass-radius relation of quark stars
2
Tolerance 10-3
Tolerance 10-4
1.8
1.6
Mass (M)
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0
1
2
3
4
5
6
Radius (km)
7
8
9
10
Figure 5.9: Comparison of a programme run with a precision of 10−3 and 10−4 for the model using three
massless quarks to two-loop running.
For low mass quark star models there is a power law behaviour R ∝ M 3 , as was expected from
the analysis of section 5.1.1. For larger mass the gradient steepens, until a maximum is reached when
dM/dr = ∞. After this point r decreases as M still increases. The maximum stable model is reached
when dM/dr = 0. After this point no stable solutions can be found using the equation of state of the
model. A list of turning points can be found in table 5.1. Note the change in central pressure between
maximum radius and maximum mass. For quark stars with small masses a slight increase in central
pressure makes the star much more massive, while for larger masses a much larger change in central
pressure is needed for a similar change in mass. This can be understood from the discussion in section
5.1.1. For low mass models gravity is unimportant and the matter behaves as an incompressible
medium. At larger masses, the effect of gravity tends to compress the matter and to let it behave more
like degenerate matter under gravity.
As is very clear from this plot, the scale Λ plays a critical role in the maximum mass. Both the
maximum mass and maximum radius are ∝ 1.5 times larger. Also the internal pressure is much lower.
This makes sense as a larger scaling of Λ with µ leads to a lower interaction strength.
Point
Rmax , Λ = 2µ
Mmax , Λ = 2µ
Rmax , Λ = 3µ
Mmax , Λ = 3µ
R(km)
6.18
5.96
12.9
12.4
M(M )
1.01
1.09
2.05
2.23
ρc (MeV/fm3 )
4.15 × 103
9.22 × 103
9.02 × 102
2.11 × 103
Table 5.1: List of maximum masses and radii for Λ = 2µ and Λ = 3µ using an equation of state of three
massless quarks.
54
5.4 Numerical solutions to quark star models using a perturbative QCD equation of state
2
Running α, ms=0
Running α and ms
1.8
1.6
Mass (M)
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0
1
2
3
4
5
6
Radius (km)
7
8
9
10
Figure 5.10: Mass-radius plot obtained by solving the structure equations 2.1 and 2.2 and the equation
of state of one massive quark flavour (equation 4.42) compared to the equation of state of three massless
quarks. Both models are calculated up to first order in α s . The coupling constant is also calculated to one
loop running. Both models use a scaling Λ = 2.
5.4.2
Quark star models using an equation of state of one massive quark flavour
In figure 5.10 the effect of giving one quark flavour a mass is shown. This equation of state is compared
to an equation of state for three massless quarks. Both equations of state contain first order loop
corrections. Comparing to the results of the previous section, second order contribution leads to
∼ 1.5 times smaller masses and radii. Inclusion of a running strange quark mass results in a change
in maximum mass and radius with a factor of about 1.5, in effect giving a similar contribution as
including the second order correction. Although I use a different equation of state as in Fraga and
Romatschke (2005), as is explained in 5.3.2, their conclusion that the mass of the strange quark should
not be neglected is still justified by my plots. However, the masses obtained are rather different. Using
my equation of state the maximum masses and radii are about half the masses and radii of Fraga and
Romatschke (2005).
These models were again calculated with a precision of 10−4 on µ and M, which also leads to
small errors on R as shown in figure 5.11. In table 5.2 a list of maximum masses and radii obtained
from figure 5.10 is given.
Point
Rmax , m s = 0
Mmax , m s = 0
Rmax , m s , 0
Rmax , m s , 0
R(km)
9.13
8.96
6.59
6.27
M(M )
1.92
1.99
1.02
1.14
ρc (MeV/fm3 )
2.01 × 103
3.48 × 103
4.00 × 103
1.01 × 104
Table 5.2: Maximum mass and radius for a model using an equation of state assuming three massless
quarks and model using an equation of state assuming one massive quark flavour, all to one loop running.
55
Chapter 5: Mass-radius relation of quark stars
1.2
Tolerance 10-3
Tolerance 10-4
Mass (M)
1
0.8
0.6
0.4
0.2
0
2.5
3
3.5
4
4.5
5
5.5
Radius (km)
6
6.5
7
Figure 5.11: Comparison of a programme run with a precision of 10−3 and 10−4 for the model using one
massive quark flavour.
When using the programme to calculate quark star models based on an equation of state with
one massive quark flavour, some models gave odd results. These results were not on the mass-radius
relation as plotted in 5.10. I have deleted these points as they are the result of an error in the numerical
calculation. This problem is discussed in section 5.2.3.
5.5 Conclusions on the mass-radius relation
The quark star models described above certainly lead to realistic predictions mass and radius. The
radii of quark stars are just below radii of neutron stars with similar masses, while also maximum
allowed masses are in the range of observations (see figure 8.1). As discussed more comprehensively
in the discussion, chapter 8, there seems no observation based constraint on mass or radius to exclude
equations of state based on perturbative QCD as equations of state for compact stars.
Unfortunately, the allowed mass and radius for a quark star is very dependent to the equation of
state. The difference in the mass-radius relation for a scaling Λ = 2µ and Λ = 3µ is about a factor
1.5. Also the mass-radius relation based on perturbative QCD up to one loop corrections of three
massless quarks is very different from the mass-radius relation from perturbative QCD to two loop
corrections. Apparently the two loop correction is still very significant. This leads to the idea that also
the three loop corrections might have strong influence on the equation of state and the models based
in it. Unfortunately, up to today the three loop correction has not been calculated. As the two loop
corrections where calculated now some thirty years ago, apparently the calculations of the three loop
corrections are much more complex.
An important observation of the last section is that ignoring the strange quark mass is not justifiable by calculations. Inclusion of the mass leads to very different mass-radius relations, and there is
no reason to assume that this situation will be different up to two loop corrections.
Calculations based on phenomological models, such as the MIT Bag model and the NJL model
56
5.5 Conclusions on the mass-radius relation
also give similar mass-radius relations. However, these mass-radius relations are also strongly dependent on the free parameters in the theory. The MIT Bag relation strongly depends on a bag pressure
that can only be fitted at low density. Mass-radius relations based on an NJL equation of state depend
on the four fermion coupling strength, again a parameter that can only be fitted at low density.
Finally, as shown by Rüster and Rischke (2004), inclusion of colour superconductivity might
be a strong influence on the mass-radius relations. For a strong coupling in the NJL model colour
superconductivity occurs in the whole region of weakly interacting quark matter, leading to very
different mass-radius relations. Their result is in very strong contradiction with Buballa (2005), who
finds only a few percent change in the mass-radius relation due to colour superconductivity.
57
CHAPTER 6
Theory of superconductivity
In this chapter the theory of superconductivity will be reviewed. Superconductivity can arise when
an attractive interaction is available that can bind two fermions. Cooper (1956) showed that such an
interaction can, for any interaction strength, always lower the ground state. This can be explained by
realising that the creation of Cooper pairs lowers the Fermi levels of the system. The chemical potentials of degenerate fermions are about the Fermi energy: µ ' F , thereby creating an energy gain. This
allows even very weak interactions to create these pairs and thereby superfluid or superconducting
behaviour.
In the first section we will see the consequences when the attractive potential is caused by electromagnetic interactions. The collective behaviour will strongly influence the electromagnetic fields
in the system, causing superconductivity. We will also explore the effect when there is an attractive
potential between neutral particles, causing in superfluidity.
The second section is about colour superconductivity. We will see that the strong force can also
create an attractive potential between fermions, thereby allowing the fermions to form Cooper pairs.
Due to the more complicated nature of the strong force different types of Cooper pairs will form,
creating a multitude of colour superconducting states. We will explore the effects of these states on
electromagnetic fields and rotation.
6.1 Superconductivity in neutron matter
The microscopic theory for superconductivity in the J = 0 channel is the BCS theory (Bardeen et al.,
1957). This theory applies to low temperature superconductors and probably proton-superconductivity
as expected in neutron stars. The theory requires an attractive interaction that allows the formation of
Cooper pairs. Pairs are formed by fermions with opposite momentum near the Fermi surface, the total
pair momentum is thus zero. This results in an isotropic condensate of Fermi pairs.
In Earth superconductors these pairs are electrons, and the attraction that binds them is a phononexchange interaction. Qualitatively this effect can be understood by realising that at a microscopic
level a lattice is not a homogeneous entity. When an electron passes through a lattice, it attracts
positive nuclei through the electromagnetic interaction. Due to the much larger mass of these nuclei,
they respond much more slowly on the movements of the electron, and are still somewhat displaced
when the electron is long gone. This creates a slight positive charge in the wake of a moving electron,
allowing it to bind to another electron.
Since this interaction is very weak, binding occurs on very long length scales. In Earth supercon-
59
Chapter 6: Theory of superconductivity
ductors, these length scales often reach many times the average electron separations distance. This
results in a collective behaviour since the wave functions of the pairs will be strongly overlapping.
The amount of overlapping can be described by a length scale, the coherence length. The coherence
length describes the average size of the wave function of electrons in the superconducting state.
One good example of such collective behaviour is the Meissner effect. This effect describes the
tendency of a superconductor to drive out any magnetic field. In low temperature superconductors
the coherence is such that there only exist two situations when a magnetic field is applied. For low
fields, the superconductor will perform perfect diamagnetism and expel the complete field. At a
certain material dependent field strength superconductivity is destroyed and the field can pass the now
normal conductor. This behaviour is known as type I superconductivity.
6.1.1
Type I superconductivity
It can be understood by a length scale comparison. The first length scale, the coherence length,
is already discussed. Another important length
scale is the penetration depth. This length scale
give the characteristic distance it takes for the superconductor to lower the field to 1/e of its vacuum value. Usually, this is only a very short distance, stretching a few tens of lattice sites. Using as notation λ(T ) as the penetration depth and
ξ(T ) as the coherence length, type I superconductivity arises when (Tinkham, 1996)
κ=
λ(T )
1
< √
ξ(T )
2
(6.1)
Figure 6.1: When an infinite slab of type I superconducting material blocks an electromagnetic
field, it is energetically more favourable to allow
the field to pass. For this, formation of domains
where H = Hc will form, destroying superconductivity there but allowing the field to pass (image
from Tinkham, 1996).
Otherwise, formation of flux tubes is more energetically favourable than completely expelling
the magnetic field.
Yet, it is possible for magnetic fields to penetrate type I superconductors by forming nonsuperconducting domains, as shown in figure 6.1
for an infinite two-dimensional slab of superconducting material. These domains differ from the fluxoids present in type II in that they are larger and contain a larger magnetic flux that is not quantised
(see section 6.1.2). These domain structures form in non-trivial geometries. In a sphere, for example,
the orthogonal field on the poles is of a much higher value than on the equator, where it drops to
zero. This can lead to the situation where at the poles the critical field is reached, while at the equator
naive analysis suggests that the magnetic field can be expelled. In this situation various domains of a
non-superconducting state will form, allowing the field to pass.
To understand this behaviour, we need to study the energetics of this situation. Following the
outline of Tinkham (1996), we need to compare the energy of a normal state with magnetic field to a
superconducting state with an expelled field. The energy of a sample of volume V in the normal state
is
µ0 H 2
Fn = V fn0 + V
+ Vext µ0 H 2
(6.2)
2
60
6.1 Superconductivity in neutron matter
with fn0 the free energy density of the normal state without a field, Vext the volume of the field and H
the auxiliary magnetic field. For the superconducting state the free energy is given by
F s = V f s0 + Vext
µ0 H 2
2
(6.3)
with f s0 the free energy density of the superconducting state. Combining these yields
Fn − F s = V( fn0 − f s0 ) + V
µ0 Hc2
µ0 H 2
µ0 H 2
=V
+V
2
2
2
(6.4)
since fn0 − f s0 = µ0 Hc2 /2.
Now we apply this result to a sphere. Outside the sphere the field satisfies the free Maxwell
equations
∇ · B = ∇ × B = ∇2 B = 0
(6.5)
where B → H for r → ∞.
Inside the sphere we have two possibilities. If H Hc the field will be completely expelled. Yet
for stronger field different behaviour can occur. If H < Hc then B = 0 inside the superconducting
medium. This leads to the extra boundary condition
B⊥ = 0 at r = R
with R the radius of the sphere. The solution to this problem is then
µ0 HR3 cos θ B = µ0 H +
∇
2
r2
(6.6)
with θ the polar angle with respect to H. From this equation we can obtain the tangential component
3
B// = µ0 H sin θ
2
(6.7)
Now B// > µ0 H for 0.73 < θ < 2.41. This means that while the average field can still be below
Hc , for parts of the sphere H > Hc . As soon as H = 2/3Hc the field at the equator will reach Hc .
From that moment, certain regions of the sphere have to be in the normal state. The coexistence of a
superconducting state with non-superconducting domains can thus occur over the range of
2
Hc < H < Hc
3
(6.8)
As shown in Tinkham (1996) the magnetisation increases linearly from zero to Hc as the applied field
increases from 2/3Hc to Hc .
6.1.2
Type II superconductivity
√
When κ > 1/ 2, with κ as in equation 6.1, a very different behaviour occurs. The transition can
be derived when realising that there actually exist two critical fields. One is the critical field Hc as
discussed in the previous section, the other is a critical field Hc2 that indicates that stable vortices can
penetrate the material.
The second field can be obtained using the linearised Ginzburg-Landau equation. Again, the
derivation closely follows Tinkham (1996). This equation reads
!2
2πA
ψ
ψ = −2m∗ αψ ≡
−i∇ −
(6.9)
Φ0
ξ(T )
61
Chapter 6: Theory of superconductivity
In this equation m∗ is the effective mass, α the electromagnetic coupling constant, A the magnetic
vector potential and Φ0 a quantised amount of magnetic flux, as will be shown in equation 6.20. The
last equivalence is in the definition of the characteristic length ξ(T )
ξ(T )2 =
1
2m∗ |α(T )|
(6.10)
while ψ is the Ginzburg-Landau (GL hereafter) complex order parameter, describing the fraction of
the material in the superconducting state.
Using this framework, we can derive when nucleation of superconductivity arises in a bulk sample.
The field is taken parallel to the z axis, with a possible gauge choice
Ay = Hx
with all other components of A equal to zero. We furthermore choose 0 = µ−1
0 = 4π. Then we can
expand the left hand side of equation 6.9 to obtain
"
2! #
(4π)i
∂
1
2
2 H
−∇ +
Hx + 2π
(6.11)
x2 ψ = 2 ψ
Φ0
∂y
Φ0
ξ
with Φ0 = 1/2e. This solution only depends on x, so we make an Ansatz of the form
ψ = eiky y eikz z f (x)
(6.12)
which we insert into equation 6.11 to obtain
− f (x) + 2µ0 π
00
2
H
Φ0
!2
with
x0 =
!
1
2
(x − x0 ) f = 2 − kz f
ξ
(6.13)
ky Φ0
(4π)H
(6.14)
2
When we multiply equation 6.13 with 1/2m∗ , we get the Schrödinger equation for a particle of
effective mass m∗ in a harmonic oscillator with force constant
!
!2
2π2 H
m ∗ Φ0
This equation can be solved an gives energies
!
!
!
1 2eH
1
n = n +
≡ n + ωc
2 m∗ c
2
(6.15)
These solutions have to be equated with the right hand side of equation 6.13. From this, the field H is
given by
r
!
1
1 Φ0
2
H=
−
k
(6.16)
z
4π2 2n + 1 ξ2
The maximum of this function is found for n = 0 and kz = 0, then equation 6.16 yields
Hc2 =
62
√
4πλ2 Hc2
Φ0
=
= 2κHc
2πξ0
φ0
(6.17)
6.2 Colour superconductivity
√
From equation 6.17 it can be seen that for κ > 1/ 2 the field Hc2 > Hc . This leads to the possibility
of nucleating superconductivity before it is energetically favourable to drive out the entire magnetic
field. In the step from equation 6.16 to equation 6.17 we have used the definitions
Φ0
√
2 2Hc (T )λeff (T )
√
2
λeff (T ) 2 2πHc (T )λeff (T )
=
κ=
ξ(T )
Φ0
ξ(T ) =
(6.18)
(6.19)
For fields H : Hc < H < Hc2 it is possible to have a coexisting magnetic field and superconductor.
The field penetrates the superconductor in fluxoids. A peculiarity is that these fluxoids can contain
only a very specific amount of field
2π
Φ0 = n ∗ = nΦ0
(6.20)
e
Here e∗ is the effective charge of the pair, which is ideally 2e for a pair of two particles with charge e.
As it is energetically favourable to form as many fluxoids as possible, n will often be 1.
The quantisation of flux can be understood from the Ginzburg-Landau idea of having a singe
complex GL order parameter ψ. In order for ψ to be single-valued, only a phase change for circling
the fluxoid of 2πn is allowed.
From this it can also be seen that these fluxoids are topological defects of the smooth superconductor. It is not possible to bring such a defect into being by a continuous gauge transformation of the
fields. The only way of inserting a fluxoid is by creating one at the edge of the superconductor.
6.2 Colour superconductivity
When a material becomes superconducting to electromagnetic fields, the U(1) symmetry of electromagnetism is globally broken due to electron pairing. This results in the photons acquiring an effective
mass, damping the electromagnetic field at very short length scales. At the same time the pairing of
electrons in Cooper pairs puts them in a bosonic ground state, generating a sort of electronic BoseEinstein condensate in the material. The result is that any attempt to apply an electromagnetic field
will be cancelled by the system, in effect allowing an electric current to flow without any resistance
whilst expelling all magnetic fields. For this mechanism to happen an attractive channel for the pairing
particles is needed. In low-temperature superconductors phonon exchange is providing the required
mechanism that creates an attractive potential for electrons in the superconductor.
A similar mechanism can also happen to fermions carrying a colour charge. The 3 ⊗ 3 representation of the SU(3) symmetry group of quark interaction can be decomposed in 6 ⊕ 3̄, where the 3̄
channel is attractive. This can be seen by realising that single gluon exchange is attractive between
two quarks antisymmetric on colour. This is the case for the 3̄ colour channel (Rajagopal and Wilczek,
2000). That strong binding occurs in this channel can be understood by realising that for strong pairing to happen the wave function must be symmetric in r. As 3̄ is antisymmetric in colour for states in
this representation a wave function can be symmetric in position space, leading to a bound state. For
stronger coupling constants the there can also be a contribution of the instanton interaction, which is
also attractive in this channel.
Since a quark condensate typically consists of more than one particle species, and the charge is a
root from the 3 representation of the SU(3) symmetry group, it shows a more complex behaviour than
electromagnetic superconductivity. There are quite a few distinct phases present, which mainly differ
in the particles participating in the pairing.
63
Chapter 6: Theory of superconductivity
6.2.1
Colour-Flavour Locked phase
At very high densities well above the strange mass, weak equilibrium will guarantee that all three low
mass quark flavours will be present. In that case, the lowest energy attainable is achieved by pairing
all three quarks in the so called colour-flavour locked phase. This condensate is of the following form
(Rajagopal and Wilczek, 2000)
D
E
D
E
bβ
aα bβ
2 αβA
ψaα
abA
(6.21)
iL ψ jL ab = − ψiR ψ jR ab = ∆(p )
Here {a, b}, {α, β} and {i, j} are indices for respectively spin, colour and flavour. In the last term the
Levi-Civita tensor clearly links colour and flavour, hence the
colour-flavour
locked. The LeviE
D name
bβ
ψ
antisymmetric
in both colour
Civita tensor is an antisymmetric tensor, making the states ψaα
ab
iL jL
and flavour indices. The binding energy of these pairs is very high, with expected superconducting
gaps of the order of 50 − 100 MeV.
Consequence of this locking is a symmetry breaking similar to that in the BCS theory of superconductivity. There is a non-zero expectation value for condensate pairs, which break the symmetry
of massless QCD to a much smaller group (Rajagopal and Wilczek, 2000)
SU(3)colour + SU(3)L + SU(3)R + U(1)B → SU(3)colour+L+R × Z2
(6.22)
An interesting question in this context is the effect of symmetry breaking on the symmetry group of
electromagnetism. Electromagnetism couples to the different quarks with different field strengths ( 23 e
for u and − 13 e for d and s), so the general U(1) symmetry of electromagnetism is definitely broken by
quark cooper pairing. However, like in the analogous theory of electro-weak symmetry breaking, a
linear combination respecting U(1) symmetry survives. The symmetry is (Alford et al., 2000)
e = Q + ηT 8
Q
(6.23)
√
e is the charge in the surviving symmetry and Q and T 8 are
with η = 1/ 3. Here, Q
 2
 3 0

Q = e  0 − 13

0 0
0
0
− 31





 1 0 0

T 8 =  0 1 0

0 0 −2




e
By this construction, all Cooper pairs carry no charge in this new symmetry. As the condensate is Q
neutral, this new symmetry will be unbroken. Part of the electromagnetic field will be expelled by
e field. For a CFL-phase with no
the Meissner effect, but a large fraction of the field can pass as a Q
electron contribution this would lead to the diffraction of light waves similar to glass or diamond.
6.2.2
Two-quark superconducting phase
At lower energies the comparatively large mass of the strange quark will spoil the symmetries allowing
CFL-superconductivity. As weak equilibrium does not guarantee equal number density of u, d and
s in this case, the flavour symmetry is not possible. Still, given their small mass, assuming weak
equilibrium between u and d is reasonable. In this case there exists the possibility of the u and d
quarks pairing in a two species superconducting phase, or the 2SC-phase.
In the 2SC phase, pairing of flavour singlets (ud − du) occurs. This pairing breaks the full SU(3)
symmetry of QCD down to a SU(2) subgroup, leaving one colour unpaired. Although in the case of
massless quarks the CFL-phase is energetically favoured, for massive quarks the 2SC-phase can be the
64
6.2 Colour superconductivity
ground state at certain energies. The strong interaction of quarks leads to a large gap ∼ 50 − 100 MeV
also for the 2SC-phase (Alford et al., 2000, Buballa, 2005)
There are a few important distinctions between the CFL phase and the 2SC phase. This phase
leaves the global flavour symmetry SU(2)L ⊗ SU(2)R intact, and also the global U(1) symmetry of
rotation survives. Therefore, the 2SC-phase is not a superfluid. Again, the U(1) symmetry is a linear
combination of the photon with one of the eight gluons. Furthermore, this phase respects chiral
symmetry. Since this symmetry is broken at low densities, there has to be a phase transition from low
densities to the 2SC phase. Increasing the density further, chiral symmetry will be broken again by
the CFL phase, requiring another phase transition. In Rajagopal and Wilczek (2000) it is argued that
this phase transition has to be first order, as the order parameter of the husi and hsdi to be nonzero
requires gaps ∆us and ∆ds greater than m2s /2µ.
In that same article it is calculated that at µ = 400 − 500 MeV, an energy scale relevant for quark
stars, both the CFL phase and 2SC phase are possible, depending on the interaction strength. The
strength of the interaction is currently not known precise enough to predict the favoured state.
6.2.3
One flavour pairing
It is also possible that one flavour will pair with itself. In the 2SC phase, for example, the strange
quark is left unpaired from the ud − du condensate. However, this does not forbid the existence of
a hssi condensate. In this case, the pair needs an angular momentum J = 1, since pairing if similar
particles with angular momentum J = 0 would violate the exclusion principle. This results in a pairing
with a small gap compared to the s-wave colour superconducting states like CFL and 2SC, estimated
to be of the order of a few hundred keV.
6.2.4
Other colour superconducting phases and the quantum chromodynamics
phase diagram
The analysis of the CFL phase assumes a symmetry between the three participating quarks. However,
the strange quark does possess a considerable mass, possibly breaking the symmetry of CFL. At lower
energies, the mass of the strange quark might prevent pairing of two quarks with opposite momenta
due to unequal Fermi momenta. One of the possibilities is the existence of a gapless phase. However,
studies (Alford and Rajagopal, 2006, Rajagopal and Sharma, 2006, and references therein) has shown
that the gapless phase does not represent the state of lowest free energy, and that the ground state
must be formed by a pairing of particles into Cooper pairs with a net momentum. This phase was
first studied by Larkin and Ovchinnikov (1964) and Fulde and Ferrell (1964) and is known as the
LOFF-phase. This results in breaking translational invariance, hence the name crystalline colour
superconductivity.
Another possibility is the formation of a Kaon condensate of K 0 particles formed by binding two
Cooper pairs into a particle like excitation with the quantum numbers of a Kaon (see Alford and
Rajagopal, 2006, and references therein). According to this reference the Kaon condensate is easily
destroyed by instanton effect. This conclusion is argued by Warringa (2006) who performed a detailed
calculation of the instanton effect on the Kaon condensate and does find a stable CFL-K0 phase at low
temperature.
Using the NJL effective model it is possible to calculate the QCD phase diagram (figure 6.2). In
the NJL effective theory the one-gluon exchange interaction is replaced by an effective four-fermion
interaction. This simplification allows computation of various parameters, like the order parameter
of the superconducting phases. The strength of the four-fermion interaction is not well determined,
65
Chapter 6: Theory of superconductivity
leading to a significant uncertainty in this diagram. However, the general structure of figure 6.2 does
not depend on this.
The existence of the different phases can also
be understood from a physical perspective. At
zero temperature and density quarks are confined and chiral symmetry is broken. At very
high temperatures both these symmetries are restored, making phase transition necessary. At
very high densities the CFL-phase is expected
to be the ground state of matter. This condensate
also possesses a broken chiral symmetry, since
colour and flavours are linked. Yet, confinement
is lifted, and thereby the chiral symmetry breaking mechanism of low density is restored. This
leads to the possibility of a phase transformation
from confined matter to a chiral symmetric quark
condensate.
.
66
Figure 6.2: The QCD phase diagram. At high temperatures formation of a quark-gluon plasma occurs. At high densities neutral quark matter in one
of the colour superconducting states is expected
(image from Formal, 2007)
CHAPTER 7
Magnetic fields in hybrid stars and
quark stars
In the previous chapter we have discussed the properties of electromagnetically and colour superconducting matter. Now, we will explore the consequences for the magnetic properties of compact stars
containing quark matter. As is found by Alford et al. (2000) deconfined quark matter responds very
differently to electromagnetic fields than normal matter. In more recent work (Ferrer and de la Incera,
2006a,b) it is even suggested that colour superconducting quark matter may act as a paramagnet. This
in sharp contrast to the perfect diamagnetism present in electromagnetic superconductors.
7.1 Neutron star magnetic behaviour
Superconductivity has a strong influence on the magnetic properties of neutron matter. One of the
main characteristics of a neutron star is its stable field. It is generally accepted that magnetic fields
of a pulsar do not decay during its lifetime. Superconducting matter allows field lines to decay only
via moving its flux tube to the surface, where the magnetic field line can reconnect and release its
energy. This creates a very stable field, especially when combined with the electronic component.
This component is not in a superconducting sate but has very good conducting properties, generating
a strong Lorentz force on any moving field line (see section 1.2.3 for a discussion).
7.1.1
Fluxoid pinning
For a long time, the assumption has been made that the superconducting medium of a neutron star is
of type II. For fields H : Hc < H < Hc2 it is then energetically favourable to allow a field to exist
in a superconductor by forming fluxoids. As these fluxoids are tiny regions of non-superconductivity
and thus at that region the ground state energy is lifted by the superconducting gap. To minimise
the energy loss, fluxoids will pin to sites where this gap has a minimal value. A superconductor is
calculated assuming a smooth background, the defects therein are the pinning sites. In order to be
able to pin to such a defect, it has to be larger than the correlation length ξ. For proton matter ξ is very
small (Anderson et al., 1982)
vF
ξ=
(7.1)
∆
where vF is the Fermi velocity and ∆ is the superconducting gap. Using this relation and equation
6.1 leads to the following condition on the penetration depth of proton matter not to be a type II
67
Chapter 7: Magnetic fields in hybrid stars and quark stars
superconductor:
1 vF
λ(T ) < √
∼ 1 fm
2∆
(7.2)
For an assumed vF ∼ 100 MeV and ∆ ∼ 0.5 MeV. The magnitude of λ is typically of the order of the
London penetration depth
1
4πe2 n s
=
(7.3)
m∗
λ2
with n s the superconductor carrier density and m∗ the effective mass of the superconductor carrier.
Assumed values for λ are of the order of 80 fm, in which case type II superconductivity is favoured.
According to Buckley et al. (2004) the self-interaction changes this picture and makes type I superconductivity favoured. In Alford et al. (2005), Jones (2006) it is shown that the suggested interparticle
interaction is too weak to for type I superconductivity.
Apart from the superconducting proton component a neutron star is also thought to possess a superfluid neutron liquid. A characteristic property of superfluidity is that rotation is accommodated by
the formation of vortices containing all angular momentum. The behaviour of a superfluid to rotation
is very similar to that of a superconductor to electromagnetic fields. This is not very strange, as the
symmetry of two dimensional rotation is also a U(1) symmetry, just as the symmetry of electromagnetism is a U(1) symmetry. Strong evidence for the presence of a superfluid containing vortices is
based on glitches, which are thought to occur as a result of vortex pinning in the lattice near the surface. See section 1.1.1 for a discussion about glitches or section 7.2.1 for a detailed discussion of the
theory of glitches assuming vortex pinning.
It is possible to calculate the fluxoid and vortex density inside a neutron star. Superfluid vortices
are quantised in angular momentum, and line up parallel to the rotation axis. The angular momentum
per vortex is
I
2π
v × dl =
n
(7.4)
m
C
were n = 1 is the state of lowest energy. Using this one can derive a vortex density (Hsu, 1999)
nS V = 104 /P(sec) cm−2
(7.5)
leading to 104 vortices per cm−2 for most pulsars and 108 per cm−2 for the fastest spinning millisecond pulsar.
Using the relation for the flux quantum (equation 6.20) and realising that flux is quantised per
proton Cooper pair of charge +2e we can also derive a flux tube density
nFV = 107 B cm−2
(7.6)
Assuming typical magnetic fields ∼ 108 T this leads to a much higher density of flux tubes than
rotational vortices. The flux tube lattice is expected to be much less regular than that of the vortices
(Link, 2003, 2006)
Realising that neutrons have an associated magnetic moment and recalling that flux tubes are often
created on defects, it is easy to imagine that flux tubes are strongly coupled to these vortices. In Link
(2006) it is calculated that the associated energy per vortex-flux tube pairing of ±5 MeV, depending
on whether the vortex and tube are parallel or anti-parallel. This is a strong pinning, making it nearly
impossible for the vortices and fluxoids to move independently from each other, whether this is regular
motion or slow vortex creep.
68
7.1 Neutron star magnetic behaviour
Flux pinning consequences
This pinning has two consequences. First, it relates the magnetic field and rotation period. As the
star spins down, it needs less and less vortices to contain angular momentum. These vortices can only
disappear through the surface, taking the pinned fluxoids with them. This effect is known for a long
time and leads to a weakly time dependent magnetic field (Bhattacharya and Srinivasan, 1995)
B(t) ∝ Nv = 2 × 1016 P−1
(7.7)
with P the period in seconds and Nv the number of vortices, to which the magnetic field is proportional
if the fluxoids are perfectly pinned to these vortices.
For a solitary pulsar spinning down this would lead to a magnetic field decay of (Bhattacharya
and Srinivasan, 1995)
t −1/4
B(t) ∝
(7.8)
τ
with τ the characteristic spin-down timescale, typically 106 − 107 years. This slow change is not
ruled out by observations. The effect only becomes appreciable when mass transfer and tidal forces
can seriously influence the spin. In low mass X-ray binaries a neutron stars with spins of more than
1000 s have been found. This would reduce the magnetic field by a factor of 104 or more, and can be
used to explain the low magnetic fields of millisecond pulsars thought to originate from these objects.
Yet, this result is contradicted by the finding that these slowly rotating neutron stars still have a strong
field ∼ 108 T (Coburn et al., 2006, more discussion in section 1.1.1)
Another effect of this pinning is the strong coupling of the neutron superfluid to the proton superconductor. This proton superconductor is via the magnetic field coupled strongly to the surface. This
in effect makes the neutron star a rigid rotator. This has severe implications on higher order perturbations of the rotation. In Link (2003, 2006) it is calculated that in this case the frequency of precession
is of the same order as the rotation period, or faster.
For a rigid biaxial body of oblateness the precession frequency ω̄ in units of ω is ω̄ = ω.
Models of neutron stars indicate an oblateness ∼ 106 , which is supported by observation of Stairs
et al. (2000). If, however, the neutron liquid would couple perfectly effective to the body via vortex
pinning, the precession frequency would change to
ω̄ = ω +
Lp
Ib ω
(7.9)
In this equation L p is the angular momentum of the pinned superfluid and Ib the moment of inertia of
the rest of the object. Since L p ' I p ω and therefore L p /Ib ω ' I p /Ib , this will be the dominant
contribution to precession frequency. Using this relation, and realising that most of the inertia will
be contained in the neutron superfluid, estimates for the precession frequency yield ω̄ = I p /Ib ≈ 10,
much higher than observations. This result was first obtained in Link (2003). In a later article (Link,
2006) it is shown that also imperfect pinning, known as vortex creep, is inadequate to explain the
low precession periods observed. From these articles it seems that a neutron superfluid and a type II
superconductor cannot coexist. Instead a type I superconductor is proposed. I will first discuss the
possibility of a colour superconducting compact star core, coming back to the possibility of type I
superconductivity in section 7.2.
69
Chapter 7: Magnetic fields in hybrid stars and quark stars
7.1.2
Quark matter influence on magnetic fields
Colour superconductivity and electromagnetism
A possible solution for the precession paradox discussed above is the presence of quark matter in the
interior of a compact star. As mentioned in section 6.2, breaking colour symmetry only shields the
electromagnetic field marginally. More important, in the CFL phase the presence of an electromagnetic field does not lead to the formation of fluxoids. Rotation however, does lead to vortex formation
in the CFL phase.
In the 2SC phase fluxoids might be present (Iida and Baym, 2002), but the density will be much
lower as the field is reduced by only 1/40 (Alford et al., 2000). When rotating matter in the 2SC phase
something more interesting occurs, as rotation of this medium can create a London magnetic field
(Iida and Baym, 2002)
√ 
!
!
 3  1 s
µ/3
8


|B | ∼ 0.015 
T
(7.10)

g8 Prot 300 MeV
which is negligible with the surface magnetic field of a neutron star. While an interesting effect it is
probably safe to neglect it.
Stability of magnetic fields in quark matter
While quark matter is not an electromagnetic superconductor, in Alford et al. (2000) it is found that it
is an extremely good conductor. Ohmic decay times of the magnetic field are in excess of the Hubble
time, and therefore easily in range of observations.
A drawback of this view is that a compact star containing quark matter does not have a natural
decay of the magnetic field in a binary, as happens with a compact star containing a type II proton
superconductor and a neutron superfluid (equation 7.7).
Precession of neutron stars with quark matter
Using the results from equation 7.9, but allowing for a quark matter component, we can calculate
precession rates for compact stars containing quark matter. We will use the assumption that quark
matter and neutron matter are neatly separated into different regions. This allows us to approximate
the relative size compared to the star of the volume were pinning occurs. This assumption is not
unreasonable, as it is thought that there might be a sharp transition between hadronic matter and
quark matter. Furthermore, I will assume a uniform density, something that is certainly not the case
but it does give an upper bound as a useful first approximation.
Suppose our star consists of a quark core up to radius a, a proton superconductor mixed with a
neutron superfluid to radius b and an crystalline iron crust to radius c, as in figure 7.1. The surface
is accounted for in the oblateness parameter, and is not needed specifically in our calculation. The
volumes of the quark core Vq and hadronic mantle Vh are
4
Vq = πa3
3
4 3
Vh = π(b − a3 )
3
(7.11)
(7.12)
so assuming uniform density the fraction of the total hadronic mass to total quark mass will be
Mh b3
=
−1
Mq a3
70
(7.13)
7.1 Neutron star magnetic behaviour
Figure 7.1: Three-layer model of a compact star. Up to radius a a fluxoid-free region is assumed. Between
radius a and b fluxoids are present, the iron crust lattice ranges from radius b to c.
If we assume that only the hadronic component couples to the magnetic field via fluxoid pinning,
the precession equation 7.9 changes to
ω̄ = ω +
!
L p b3
−
1
Ib ω a3
(7.14)
thus only the angular momentum contained in the hadronic mantle contributes to the precession.
Using this equation and observed precession rates we can derive a maximum mass fraction of a layer
of superfluid and type II superconducting matter. Assume that the weight of the crust is negligible, and
that ω̄ ∼ 10−6 ω. Then the mass fraction of the superfluid and type II superconducting component can
be no more than 10−6 . Allowing for vortex creep increases this number but will not allow a significant
fraction of the star to be in a type II superconducting state (Link, 2006).
This tiny fraction suggests the conclusion that, if type I superconductivity does not occur, quark
matter has to exist up to the region where protons and neutrons start forming a lattice, and that nowhere
a neutron superfluid fluid can coexists with a type II proton superconductor. This would mean that
quark matter is stable from chemical potentials above the chemical potential of the crust lattice.
This all leads to the conclusion that free precession excludes the possibility of type II superconductivity. Type I superconductivity is still possible, with a damping period of about 100 years (Link,
2006). The presence of quark matter also allows precession. As there is very little interaction between
a quark matter core and a hadronic crust, much longer damping periods can be expected in this case.
71
Chapter 7: Magnetic fields in hybrid stars and quark stars
7.2 Consequences of quark matter for pulsar timing observations
7.2.1
Glitches
Due to electromagnetic braking, the rotational frequency of a pulsar decreases in time. The energy and
angular momentum are carried away by the electromagnetic radiation. This mechanism is discussed
in more detail in section 1.1.1. Also discussed in that section is that sometimes there is a sudden
increase in rotational frequency, a glitch. Glitches are most commonly observed in young pulsars
and are thought to originate from an angular momentum transfer between core and crust. The theory
is that as the star spins down, the superfluid core needs to spin down as well, due to the interaction
between the proton superconductor and the neutron superfluid. For a superfluid to spin down, vortices
have to be destroyed. This can only happen at the surface, so the vortices move outwards as the star
spins down.
However, in the boundary layer between core and crust there is a coexistence of a crust lattice and
superfluid neutrons. The vortices tend to pin to lattice sites, as the interactions between neutrons are
very strong and the correlation length is very small. This pinning prevents the vortices from moving
outwards so a buildup of angular momentum occurs.
A glitch happens as a result of a massive unpinning. The released angular momentum of all the
destroyed vortices is transferred to the crust, which angular momentum increases. As only the crust
can be observed from Earth it looks like the compact star suddenly spins up.
Since the above described mechanism for glitches has nothing to do with superconductivity, it is
still valid when one assumes a type I superconducting proton component. However, as was showed
in Shaham (1977) the pinning of vortices to the crust also leads to a strong damping on precession.
A proposed solution is that in a freely precessing compact star these vortices are all unpinned. This
would mean that glitches do not happen in precessing compact stars.
It is thought that quark matter cannot coexist with normal matter, so for quark matter a different
explanation for glitches is needed. Again, we need a mechanism to store and then suddenly release
angular momentum. A mechanism to store angular momentum is the vortex creation, which also
happens in a quark superfluid. It is suggested in Alford and Rajagopal (2006), Jaikumar et al. (2006a)
that a coexistence of one of the many different colour superconducting states makes it possible to
internally store angular momentum. Especially the crystalline colour-flavour locked phase (see section
6.2.4) could play an important role. For pure quark stars it is imaginable that such a mechanism would
allow an up-spinning crust. Yet, for stars consisting of a normal crust on top of a quark matter core it
is not clear how the angular momentum transfer to the crust would occur. In this light it is probably
safe to assume that any hybrid star needs at least a small portion of superfluid neutron matter.
7.2.2
Precession
In an article by Stairs et al. (2000) strong evidence for free precession of a compact star is found.
As shown in figure 7.2(a) there are timing residuals after performing all timing corrections. In order
to test this assumption it is useful to search for any periodicity in the timing residuals. The Fourier
transform of the timing residuals can be found in figure 7.2(b). In the article different mechanisms for
the observed harmonics are discussed and rejected. This observation is therefore a strong indication
of the presence of free precession in this pulsar.
The results indicate a precession period ∼ 500 days and an amplitude α ' 3◦ (Link, 2003). Using
equation 7.9 without the damping term indicates an oblateness ∼ 10−6 , a perfectly normal value for
a compact star.
72
7.2 Consequences of quark matter for pulsar timing observations
(a) Timing residuals from PSR B1828-11
(b) Fourier transform of the timing residuals
Figure 7.2: Timing residuals and its Fourier transform from PSR B1828-11. Images from Stairs et al.
(2000)
There is some evidence for other neutron
stars showing a similar type of precession (Deshpande and Radhakrishnan, 2007, Shabanova
et al., 2001). In Stairs et al. (2000) it is also suggested that timing residuals in anomalous X-ray
pulsars can be explained by free precession.
Free precession of a compact star occurs due
to a non-spherically symmetric distortion. Most
probable such a distortion would be an explosion on the surface of the neutron star. A decaying magnetic field can cause major outbursts,
and would be a good candidate. If this is the
case, precession in the more violent compact
stars such as anomalous X-ray pulsars should be
more common. If it is possible to make an estimate of how often a distortion occurs, and how
many neutron stars undergo free precession, it is
possible to derive a damping time. This damping time would tell a lot about the equation of
state of the compact star interior as discussed in
section 7.1.2.
In the previous section was argued that
glitches should not happen in freely precessing
compact stars, as for free precession to happen
the neutron superfluid has to be unpinned from
the crust lattice. Detailed pulsar timing on precessing neutron stars could validate the assumption that precessing neutron stars cannot show
glitches. This would be a strong suggestion that
the precession model is correct.
73
CHAPTER 8
Conclusions and discussion
In the introductory chapters we have discussed many different observables of compact stars. There
we found several mechanisms to study the equation of state of matter in a compact star. The most
direct method is provided by a mass-radius determination. Furthermore, the interaction of matter and
magnetic fields dictates the behaviour of the field and thereby influences the star its rotation. Study of
precession, spindown and glitches are particular tools to access the star its magnetic properties.
In this thesis the mass-radius relation and the influence of the magnetic field were studied. The
results will be discussed in the next sections. Another way of probing the interior of a compact star
is by its cooling behaviour. I haven’t discussed this behaviour, especially as results obtained by this
method are highly degenerate. However, in order to study the interior of compact stars properly,
a combination of all mechanisms leaves the fewest degeneracy between different models. For an
observational study of these objects, a combination of all these observables is therefore preferred.
8.1 Mass-radius relations
There is still much uncertainty in the quark matter equation of state. This allows for mass-radius
relations differing by as much as a factor 1.5 for reasonable values for the renormalisation scale dependence a in Λ = aµ. Further study on QCD parameters, both theoretically and experimentally,
must be done in order to make better predictions. Especially the calculations of the quark matter
thermodynamic potential to three loop order should put more stringent constraints on the mass-radius
relation. Furthermore, detailed study of quark matter properties in astrophysical objects may provide
measurements that can help fix these parameters.
Comparing the curves with scaling Λ = aµ, a = 2, 3 to curves for neutron matter shows that
compact stars based made from quark matter are always smaller than their neutron matter counterparts.
This has to be, as quark matter is created when neutron matter is compressed to such densities that
confinement is no longer present.
The mass-radius relation compares very well to expectations based on the MIT Bag model as
discussed in chapter 5.1.1. The quark matter equation of state behaves as a nearly incompressible
medium at low densities, at higher densities it starts behaving more like ordinary degenerate matter.
Furthermore, the equation of state based on perturbative calculations (figure 5.4) only differs slightly
from non-ideal Bag models with a properly fit Bag constant.
Including one massive quark flavour, I have confirmed the result of Fraga and Romatschke (2005)
that neglecting the strange quark mass is not a minor perturbation to the final mass-radius relations.
75
Chapter 8: Conclusions and discussion
However, I have found a different equation of state for first order perturbation theory and have also
found different mass-radius relations. An analytic calculation supports my results.
Depending on the scale dependence expressed through a, quark matter is absolutely stable when
there is an energy gain for conversion from baryons to quark matter. For quark matter the pressure
vanishes at some chemical potential µcrit . For a scaling Λ = 3µ, for example, the pressure vanishes at
∼ 300 MeV. To compare it to a baryon chemical potential we have to look at the combined potential
of three quarks, 900 MeV, and compare this to the baryon chemical potential of iron, 940 MeV (Fraga
et al., 2001). In this case, quark matter will thus be absolutely stable. However, as discussed in chapter
2.2, even such stars will have a baryonic crust as the strange quark mass does not allow for neutral
quark matter. This charge must be compensated by electrons. At the surface, the the quark matter
density drops much faster to zero than the electron density can follow. This creates a considerable
electromagnetic potential at the surface of the quark star, able to support a baryonic crust.
For lower values of a quark matter is not absolutely stable and hybrid stars are expected. In order
to calculate such stars, an equilibrium in pressure and chemical potential between baryonic matter and
quark matter must be required in the model. Although this is not computationally difficult, it is an
extra source of uncertainty to the models.
The mass limit of quark stars seems not to be any more stringent for quark stars as for neutron
stars. The causality limit, stating that the speed of sound should be less than the speed of light,
puts boundaries on the equation of state. The maximally hard equation of state is = P, leading
to maximum masses ∼ 3 M . For compact stars this limit is more stringent than the limit based on
general relativity, stating that the Schwarzschild radius of an object should be smaller than the radius
of object
R&3
M
km
M
(8.1)
Otherwise, the object would be a black hole.
For our equation of state with scaling Λ = 3µ masses slightly above 2 M are possible, close to
the limiting mass based on causality. Including the strange quark results in a lower mass, but due to
the uncertainty in scaling the maximum mass of quark stars can still be close to the maximum mass
of a compact star. On the other hand, based on low energy information on the scaling constant, a
somewhat lower maximum mass is more likely.
For hybrid stars the limiting mass is similar to that of a neutron star. Although the above argument
may limit the mass of the quark core, it does not pose limits on the baryonic content. It does mean
that, if true that the maximum mass for the quark content is smaller than the maximum mass of a
compact star, the most massive compact stars must be made of at least a portion of baryonic matter.
Comparing the models with the observed masses of compact stars in binary systems as shown in
figure 8.1, none of the compact stars seem to be excluded based on equations of state discussed in this
thesis. In all of the objects in this figure there is strong evidence of the presence of a surface, indicating
compact stars and not black holes. It is puzzling why most of the compact stars in this figure seem to
cluster so close to the Chandrasekhar mass of 1.4 M . One explanation might be that the maximum
mass for compact stars is not much higher than the Chandrasekhar mass, allowing only a small mass
window in which these objects can reside. However, this result seems to be contradicted by the the
discovery of a compact star with a mass ∼ 2.1 ± 0.2 M (Nice et al., 2005). Another explanation
may lie in the formation history of binary systems, which may somehow favour the formation of light
compact stars. Yet, this scenario is in contrast with some of the proposed black hole systems, such as
Cyg X-1, which is estimated to have a mass of about 10 M (Abubekerov et al., 2004).
76
8.2 Quark matter and superconductivity
Figure 8.1: Determination of masses of compact stars in binary systems. It seems that masses close to the
minimum compact star mass M ' 1.4 M are favoured (image from Thorsett and Chakrabarty, 1999)
8.2 Quark matter and superconductivity
The presence of quark matter has important consequences for the behaviour of electromagnetism in
compact stars. While neutron matter is probably superconducting with respect to electromagnetism,
quark matter is not.
The stability of the magnetic fields is guaranteed for both types of matter. Superconductivity
of both type I and type II confine the magnetic field to domains. For field-lines to reconnect these
domains have to move through the surface, which is very difficult as the Lorentz force of the free
electrons in the compact object drive the domains back. In quark matter the extreme conductivity of
free electrons guarantees ohmic decay times of over the Hubble age, well above observational limits.
To explain the decay of the magnetic fields in millisecond pulsars additional mechanisms are
required both for baryonic matter and quark matter. A not yet understood mechanism that buries the
magnetic field seems the best candidate to describe the properties of compact stars in binary systems,
as it allows the final field strength to depend on the duration of accretion.
In order to understand glitches, there has to be a mechanism able to store angular momentum that
is quasi-periodically released to the surface. This is usually explained by the pinning of rotational
vortices of the neutron superfluid to the crystal lattice of the surface. However, in quark matter this
mechanism is not present. There might be a similar mechanism where angular momentum is stored
and released between boundaries of one of the different states of colour superconductivity. However,
77
Chapter 8: Conclusions and discussion
the exact mechanism is not yet unclear.
Observational evidence that type II superconductivity does not occur in at least some of the compact stars is quite strong. The observed precession discussed in chapter 7 is incompatible with the
presence of a type II superconductor. One solution is to assume a type I superconducting proton
medium, a possibility that is still a matter of debate. The presence of quark matter also solves the
precession problem, allowing for very long damping times on the precession.
Precession presents a very clear mechanism to study the interior of compact stars. In order to
study the validity of the models, it is good to search for compact stars that show both glitches and free
precession. The current models suggest that the combination is not possible, so if found this poses
a severe problem for these models. Furthermore, if the number of events causing free precession is
known, a statistical analysis of the phenomenon is possible. From this analysis the damping time can
be obtained. The damping time gives direct insight in the structure of the matter in the interior of a
compact star.
78
APPENDIX A
Derivation of the TOV equation
To derive the TOV-equation, it is useful to first consider the most general metric for a static, spherically
symmetric system. This metric is
ds2 = −e2α dt2 + e2β dr2 + r2 dΩ2
(A.1)
with α, β functions of r and the angular part given by
dΩ2 = dθ2 + sin2 θdφ2
(A.2)
Solutions for this metric in vacuum are the Schwarzschild solutions
!
!−1
2GM
2GM 2
2
2
dt + 1 − 2
dr2 + r2 dΩ2
ds = −c 1 − 2
c r
c r
(A.3)
For the metric inside a star a vacuum cannot be assumed. In this case a solution to this metric
from the Einstein equation with a nonzero energy momentum tensor has to be found
Gµν = 8πGT µν
(A.4)
1
Gµν = Rµν + Rgµν
2
(A.5)
with the Einstein tensor Gµν defined as
were Rµν is the Ricci tensor, contracted from the Riemann Tensor
Rµν = Rλµλν
The Riemann tensor is
ρ
ρ
ρ
(A.6)
ρ
Rλµλν = ∂µ Γνσ − ∂ν Γµσ + Γµλ Γλνσ − Γνλ Γλµσ
(A.7)
From the metric (equation A.1) the Christoffel symbols can be calculated
Γttr = ∂r α
1
Γθrθ =
r
Γrtt = e2(α−β) ∂r α
Γrθθ = −re−2β
Γrφφ = −re−2β sin2 θ
Γθφφ = sin θ cos θ
Γrrr = ∂r β
1
φ
Γrφ =
r
cos
θ
φ
Γθφ =
sin θ
79
Appendix A: Derivation of the TOV equation
All symbols not specified are zero or related by symmetry relations. Using the previous declarations,
one can calculate the Einstein tensors
1 2(α−β) 2β
e
2r∂
β
−
1
+
e
r
r2
1 = 2 2r∂r α − 1 + e2β
r
Gtt =
Grr
Gθθ = r e
2 2β
∂2r α
1
+ (∂r α) − ∂r α∂r β (∂r α − ∂r β)
r
!
2
Gφφ = sin2 θGθθ
(A.8)
We assume that the star is made of a perfect fluid. A perfect fluid is a fluid with an isotropic
pressure P and a rest-frame energy density . The energy-momentum tensor for a perfect fluid is
T µν = ((r) + P(r))Uµ Uν + P(r)gµν
(A.9)
with the four-velocity Uα defined as dxα /dτ, with τ the proper time. Searching for static solutions it
is possible to choose the four-velocity in the timelike direction and normalise to U α Uα = −1. Using
this, the four-velocity becomes
Uµ = (eα , 0, 0, 0)
(A.10)
The energy momentum tensor is then entirely diagonalised
T µν = ‰ · (e2α , e2β P, r2 P, r2 sin2 (θ)P)
(A.11)
Due to the diagonal form the Einstein equation splits into three equations
e−2β
1 −2β 2β
e
2r∂
β
−
1
+
e
= 8πG
r
r2
1 −2β e
2r∂r α − 1 + e2β = 8πGP
2
r
!
1
2
2
∂r α + (∂r α) − ∂r α∂r β (∂r α − ∂r β) = 8πGP
r
(A.12)
(A.13)
(A.14)
Equation A.12 only depends on and β. It is therefore helpful to define a new function m(r) that
depends only on r and β
1
m(r) =
(r − re2β )
(A.15)
2G
so that we can rewrite the metric (equation A.1) to
"
#−1
2Gm(r)
2
2α(r) 2
ds − e
dt + 1 −
dr2 + r2 dΩ2
(A.16)
r
and equation A.12 to
dm
= 4πr2 (A.17)
dr
This is the equation of mass conservation. Integrating this yields the total gravitational mass up to a
radius r
Z r
m(r) =
r0 r02 dr0
(A.18)
0
80
It must be, since at r = R, the outer radius of a star, the TOV metric should match the Schwarzschild
metric. Note that the volume element of the integral is not the proper spatial volume element of curved
space with a spatial metric γi j
√ 3
γd x = eβ r2 sin θdrdθdφ
Using this volume element we can calculate the total particle energy
Z r
(r0 )r02
Eparticle (r) = 4π
dr0
0 )/r)1/2
(1
−
2Gm(r
0
(A.19)
The difference Ebinding = Eparticle − m(r) is the binding energy of the star.
Rewriting equation A.13 in terms of m(r) one obtains
dα Gm(r) + 4πGr3 P
=
dr
r(r − 2Gm(r))
(A.20)
In order to solve the final equation obtained from the Einstein equation (equation A.14), it is convenient to use energy-momentum conservation: ∇µ T µν = 0. The nontrivial component is ν = r, all in all
the terms are:
∇0 T 0ν = ∂r αe2β (P + )
∇1 T 1ν = ∂r (βe2β )P + ∂r βe2β P
which combines to
dα
dP
=−
dr
dr
Using this equation to eliminate α in equation A.20 yields the TOV equation
( + P)
( + P)(Gm(r) + 4πGr3 P)
dp
=−
dr
r(r − 2Gm(r))
(A.21)
(A.22)
81
APPENDIX B
Notation
Z
N
Ω
x, t
r
M, M
T
µ
P
ρ
E
φ
ψ
τ = it
α
αs
g
S
L, L
H, H
Partition function
Particle number
Thermodynamic potential
place and time coordinates
radial coordinate
Mass and solar mass
Temperature
Chemical potential
Pressure
Energy density
Mass density for non-relativistic media
Energy
Boson field, angle or phase
Fermion field
Wick rotated time
Unspecified coupling constant
QCD coupling constant
Gluon charge
The action
Lagrangian, Lagrangian density
Hamiltonian, Hamiltonian density
Table B.1: List of symbols used in this thesis
In this thesis I will use natural units, so c = k = ~ = 1. The gravitational constant G will not be
put equal to 1. When units are present, I will use SI units.
For vectors and sums, Greek indices generally indicate a vector in 4 dimensional spacetime and
run from 0 to 4. Latin are mostly used for all other run over the three spatial dimensions or will run
over the degrees of freedom of a system. An exception to this rule is when labelling quantum numbers
of a colour conducting state. In this case a, b are the spin labels, i, j the flavour labels and α, β the
colour labels. When indices occur twice in an equation, summation over these indices is implied.
When a calligraphic notation is used, the density of a quantity is meant. So L is a Lagrangian
83
Appendix B: Notation
while L is a Lagrangian density. An exception is the energy density, for which I use the symbol
throughout this thesis.
Furthermore, some astronomic notation will be used. A symbol with a means the solar value of
that symbol. For example, L is the solar luminosity. In a similar fashion, ∗ means the value applicable
to a specific star.
When not specified, logarithms are with base e, so log ≡ ln in this thesis. I will only use log.
Path integrals are indicated with a script D. For example, the calculation of the expectation value
of operator F in path integral notation is to be understood as
R
Q R
iS [φn ]
DφF(φ)eiS [φ]
n Dφn F(φn )e
R
R
(B.1)
=
hFi =
Q
iS [φn ]
DφeiS [φ]
n Dφn e
where S [φ] is the action and φn is a decomposition of φ onto a base of orthonormal functions.
84
Acknowledgements
Hier ligt dan mijn scriptie, het resultaat van anderhalf jaar onderzoek. Hoewel ik dit onderzoek vooral
alleen gedaan heb had ik het toch nooit voor elkaar gekregen zonder de hulp van mensen om me heen.
Daarnaast heeft ook de gezellige sfeer op zowel het Anton Pannekoek-instituut van de UvA als bij de
afdeling theoretische natuurkunde van de VU bijgedragen aan mijn motivatie. Ik wil dan ook in de
eerste plaats mijn medestudenten hiervoor bedanken.
Meer in het bijzonder wil ik graag Rudy Wijnands en Daniel Boer noemen. Zij hebben me tijdens
het onderzoek begeleidt en aangezet tot het afleveren van dit werk. Met name dankzij hun kritiek
en aanmoedigingen is het me gelukt het onderzoek tot een goed einde te brengen. Vooral ook de
ongedwongen sfeer waarin ik altijd kon langskomen is me erg goed bevallen.
Daarnaast wil ik een paar medestudenten noemen. Ten eerste Nathalie Degenaar. Van haar heb
ik een stuk code gekregen wat ik als basis heb kunnen gebruiken voor mijn programma. Dit heeft
me met name in het begin erg geholpen met het wegwijs worden in de numerieke methoden en de
programmeertaal C.
Verder wil ik Daan Meerburg bedanken voor de diepgravende gesprekken die wij konden hebben
over veldentheorie, kosmologie en andere complexe theoretische onderwerpen. Jij hebt me er altijd
op gewezen dat ik secuur moet nadenken over problemen en er niet te snel overheen moet walsen.
Hoewel ik nog steeds een beetje chaotisch ben heb ik hiervan van jou veel geleerd. Ik wil je ook graag
erg veel succes wensen met je afstuderen en het verkrijgen van de talentenbeurs van de NWO. Ik heb
in ieder geval heel veel vertouwen in je.
Tot slot Yan Grange, ook jou wil ik graag bedanken voor je hulp bij programmeerproblemen die
ik ben tegengekomen. Gelukkig heb ik dit kunnen goedmaken door ook jou af en toe te helpen. Je
bent voor mij een grote hulp geweest, ik hoop dat het andersom ook het geval is. In ieder geval was
het altijd erg gezellig.
Buiten mijn studie wil ik vooral mijn ouders te bedanken voor de steun die ze me gegeven hebben,
niet alleen tijdens mijn studie. Bedankt ook dat jullie altijd interesse tonen in mijn bezigheden, hoe
weinig jullie daar soms ook mee ophebben.
Veel afleiding heb ik gehad tijdens het coachen van mijn roeiers. Hoewel niet altijd een grote bijdrage aan de productiviteit vind ik het erg leuk om te doen. Ik denk dat ik zonder externe bezigheden
niet zou kunnen studeren. Daarom iedereen die ik gecoachd heb, en in het bijzonder Ezra, Freek,
Titus, Simon en Sha, bedankt dat jullie al die tijd met mij hebben willen roeien!
85
Samenvatting
Zware sterren, sterren meer dan acht keer zo zwaar als
onze zon, exploderen aan het eind van hun leven in een
supernova-explosie. De buitenlagen van de ster worden
met grote snelheid de ruimte ingeblazen terwijl de kern
juist ineen stort tot een compacte ster. Deze compacte ster
weegt meestal iets meer dan onze zon, maar is maar zo’n
20 kilometer groot. Dit betekent dat de materie in een compacte ster zeer dicht opeengepakt moet zijn. Om een idee
te geven: als je één kubieke kilometer water tot de grootte van een suikerklontje perst, krijg je een vergelijkbare
dichtheid.
Bij zulke hoge dichtheden gaat materie zich anders gedragen. Zo zijn atoomkernen niet meer stabiel, en kunnen
de deeltjes in de atoomkernen (protonen en neutronen) zich
vrij bewegen. Het zou ook kunnen dat de deeltjes waarvan
protonen en neutronen gemaakt zijn (quarks) zich als vrije
deeltjes gaan gedragen. Om dit te begrijpen moeten we kijken naar de eigenschappen van de kracht tussen de quarks,
Figuur B.1: De kracht tussen quarks
de sterke kracht. Quarks kunnen bij lage energieën niet vrij
gedraagt zich als een soort veer. hoe
verder de quarks gescheiden worden,
voorkomen. Lage energieën komen overeen met grote afhoe groter de kracht tussen de quarks
standen. De kracht tussen quarks werkt als een soort veer,
wordt. Als de kracht te groot wordt
en wordt sterk bij grote afstanden tussen de quarks. Dit is
‘breekt’ de veer en ontstaan uit het vageı̈llustreerd in afbeelding B.1. Bij hoge energieën komen
cuüm twee nieuwe quarks (afbeelding
de ‘veertjes’ slap te staan en gaan de quarks zich gedragen
van Lippert, 2005)
als vrije deeltjes.
In zeer dichte materie moeten de deeltjes hele hoge
energieën hebben. Dit komt omdat twee quarks niet dezelfde energie kunnen hebben vanwege het
uitsluitingsprincipe. Als veel quarks erg dicht op elkaar geperst worden moet de gemiddelde quarkenergie dus toenemen. Voor zeer dichte materie kan de gemiddelde quark-energie zo hoog worden dat
de binding tussen quarks zo zwak wordt dat ze niet langer opgesloten zitten in protonen en neutronen.
In dat geval onstaan de net besproken quarksterren.
Voor mijn onderzoek heb ik gekeken naar de gevolgen van de aanwezigheid van quarkmaterie
op waarneembare eigenschappen van deze compacte sterren. Als eerste heb ik gekeken naar het
verband tussen massa en straal van deze sterren. Het blijkt dat dit verband erg anders is als de massastraalrelatie voor compacte sterren gemaakt van protonen en neutronen. Echter, voor massa’s iets
groter dan de massa van onze zon komen de diagrammen toch redelijk overeen. Dit zijn precies de
87
Samenvatting
massa’s van waargenomen compacte sterren.
Een ander gevolg is de koppeling van de kern van de compacte ster aan het magneetveld. Het blijkt
dat als de kern bestaat uit vrije quarks, er geen koppeling is tussen de kern en het magneetveld. Sinds
kort zijn er waarnemingen die aangeven dat de rotatieas van sommige compacte sterren schommelt,
dit heet precessie. We denken dat deze precessie alleen voor kan komen als er geen koppeling is
tussen de kern en het magneetveld. Mijn conclusie is dat deze precessie een veelbelovende manier is
om quarkmaterie op te sporen.
88
Bibliography
M. K. Abubekerov, E. A. Antokhina, and A. M. Cherepashchuk. Estimate of the Black-Hole Mass
and Orbital Inclination from the Radial-Velocity Curve of the X-Ray Binary Cyg X-1. Astronomy
Reports 48 550–562, 2004.
M. A. Agüeros, S. F. Anderson, B. Margon, B. Posselt, F. Haberl, W. Voges, J. Annis, D. P. Schneider,
and J. Brinkmann. Candidate Isolated Neutron Stars and Other Optically Blank X-Ray Fields
Identified from the ROSAT All-Sky and Sloan Digital Sky Surveys. AJ 131 1740–1749, 2006.
astro-ph/0511659.
C. Alcock and E. Farhi. Evaporation of strange matter in the early universe. Phys. Rev. D 32 6
1273–1279, 1985.
M. Alford, J. Berges, and K. Rajagopal. Magnetic fields within color superconducting neutron star
cores. Nuclear Physics B 571 269–284, 2000. hep-ph/9910254.
M. Alford, D. Blaschke, A. Drago, T. Klahn, G. Pagliaraand, and J. Schaffner-Bielich. Quark matter
and the masses and radii of compact stars. ArXiv Astrophysics e-prints 2006a. astro-ph/0606524.
M. Alford, G. Good, and S. Reddy. Isospin asymmetry and type-I superconductivity in neutron star
matter. Phys. Rev. C 72 5 055801–+, 2005.
M. Alford and K. Rajagopal. Color Superconductivity in Dense, but not Asymptotically Dense, Quark
Matter. ArXiv High Energy Physics - Phenomenology e-prints 2006. hep-ph/0606157.
M. G. Alford, K. Rajagopal, S. Reddy, and A. W. Steiner. Stability of strange star crusts and
strangelets. Phys. Rev. D 73 11 114016–+, 2006b. hep-ph/0604134.
J. O. Andersen and M. Strickland. Equation of state for dense QCD and quark stars. Phys. Rev. D 66
10 105001–+, 2002. hep-ph/0206196.
P. W. Anderson. Theory of Flux Creep in Hard Superconductors. Phys. Rev. Lett. 9 309–311, 1962.
P. W. Anderson, M. A. Alpar, D. Pines, and J. Shaham. The rheology of neuron stars: Vortex-line
pinning in the crust superfluid. Philos. Mag. A 45 2 227–238, 1982.
V. Baluni. Non-Abelian gauge theories of Fermi systems: Quantum-chromodynamic theory of highly
condensed matter. Phys. Rev. D 17 2092–2121, 1978.
J. Bardeen, L. N. Cooper, and J. R. Schrieffer. Theory of Superconductivity. Phys. Rev. 108 1175–
1204, 1957.
89
BIBLIOGRAPHY
G. Baym, C. Pethick, and D. Pikes. Electrical Conductivity of Neutron Star Matter. Nature 224
674–675, 1969.
D. Bhattacharya and G. Srinivasan. The magnetic fields of neutron stars and their evolution. In
W. H. G. Lewin, J. van Paradijs, and E. P. J. van den Heuvel (editors), X-ray binaries, chapter 12,
pages 495–522. Cambridge Astrophysics Series, Cambridge, MA: Cambridge University Press,
—c1995, edited by Lewin, Walter H.G.; Van Paradijs, Jan; Van den Heuvel, Edward P.J., 1995.
S. Boutloukos, M. van der Klis, D. Altamirano, M. Klein-Wolt, R. Wijnands, P. G. Jonker, and R. P.
Fender. Discovery of Twin kHz QPOs in the Peculiar X-Ray Binary Circinus X-1. ApJ 653 1435–
1444, 2006. astro-ph/0608089.
M. Buballa. NJL-model analysis of dense quark matter [review article]. Phys. Rep. 407 205–376,
2005. hep-ph/0402234.
K. B. Buckley, M. A. Metlitski, and A. R. Zhitnitsky. Neutron Stars as Type-I Superconductors.
Physical Review Letters 92 15 151102–+, 2004. astro-ph/0308148.
E. Cappellaro, R. Evans, and M. Turatto. A new determination of supernova rates and a comparison
with indicators for galactic star formation. A&A 351 459–466, 1999. astro-ph/9904225.
L. Close, R. Lenzen, J. Guirado, E. Nielsen, E. Mamajek, W. Brandner, M. Hartung, C. Lidman, and
B. Biller. A dynamical calibration of the mass-luminosity relation at very low stellar masses and
young ages. Nature 433 286–289, 2005.
W. Coburn, P. Kretschmar, I. Kreykenbohm, V. A. McBride, R. E. Rothschild, R. Staubert, and
J. Wilms. Cyclotron features in X-ray spectra of accreting pulsars. Advances in Space Research 38
2747–2751, 2006.
M. Colpi and J. C. Miller. Rotational properties of strange stars. ApJ 388 513–520, 1992.
L. N. Cooper. Bound Electron Pairs in a Degenerate Fermi Gas. Phys. Rev. 104 4 1189–1190, 1956.
A. Cumming, P. Arras, and E. Zweibel. Magnetic Field Evolution in Neutron Star Crusts Due to the
Hall Effect and Ohmic Decay. ApJ 609 999–1017, 2004. astro-ph/0402392.
N. Degenaar. Crust cooling curves of transiently accreting neutron stars. Ph.D. thesis, Universiteit
van Amsterdam, 2006.
T. DeGrand, R. L. Jaffe, K. Johnson, and J. Kiskis. Masses and other parameters of the light hadrons.
Phys. Rev. D 12 7 2060–2076, 1975.
A. A. Deshpande and V. Radhakrishnan. Further Evidence for Collimated Particle Beams from Pulsars
and Precession. ApJ 656 1038–1043, 2007. astro-ph/0609082.
Fermilabs. 1995. URL http://www.fnal.gov/pub/inquiring/matter/madeof/index.html.
L. Ferrario and D. T. Wickramasinghe. Magnetic fields and rotation in white dwarfs and neutron stars.
MNRAS 356 615–620, 2005.
E. J. Ferrer and V. de la Incera. Magnetic Fields Boosted by Gluon Vortices in Color Superconductivity. Phys. Rev. Lett. 97 12 122301–+, 2006a. hep-ph/0604136.
90
BIBLIOGRAPHY
E. J. Ferrer and V. de la Incera. Paramagnetism in color superconductivity and compact stars. ArXiv
Astrophysics e-prints 2006b. astro-ph/0611460.
D. Formal. 2007. URL http://en.wikipedia.org/wiki/QCD_matter.
E. S. Fraga. Chiral Transition and the Quark-Gluon Plasma. Brazilian Journal of Physics 36 157–163,
2006.
E. S. Fraga, R. D. Pisarski, and J. Schaffner-Bielich. Small, dense quark stars from perturbative QCD.
Phys. Rev. D 63 12 121702–+, 2001. hep-ph/0101143.
E. S. Fraga and P. Romatschke. Role of quark mass in cold and dense perturbative QCD. Phys. Rev. D
71 10 105014–+, 2005. hep-ph/0412298.
B. A. Freedman and L. D. McLerran. Fermions and gauge vector mesons at finite temperature and
density. I. Formal techniques. Phys. Rev. D 16 1130–1146, 1977a.
B. A. Freedman and L. D. McLerran. Fermions and gauge vector mesons at finite temperature and
density. II. The ground-state energy of a relativistic electron gas. Phys. Rev. D 16 1147–1168,
1977b.
B. A. Freedman and L. D. McLerran. Fermions and gauge vector mesons at finite temperature and
density. III. The ground-state energy of a relativistic quark gas. Phys. Rev. D 16 1169–1185, 1977c.
T. Fujihara, T. Inagaki, and D. Kimura. Colour superconductivity and the radius of a quark star in the
extended NJL model by using the dimensional regularization. Journal of Physics A Mathematical
General 39 6371–6375, 2006. hep-ph/0511286.
P. Fulde and R. A. Ferrell. Superconductivity in a Strong Spin-Exchange Field. Phys. Rev. 135 3A
A550–A563, 1964.
R. J. GaBany. 2005. URL http://www.cosmotography.com/.
M. Gell-Mann and K. A. Brueckner. Correlation energy of an electron gas at high density. Phys. Rev.
106 2 364–368, 1957.
N. K. Glendenning (editor). Compact stars : nuclear physics, particle physics, and general relativity.
2000.
M. Gyulassy and L. McLerran. New forms of QCD matter discovered at RHIC. Nuclear Physics A
750 30–63, 2005. nucl-th/0405013.
A. K. Harding and D. Lai. Physics of strongly magnetized neutron stars. Reports of Progress in
Physics 69 2631–2708, 2006. astro-ph/0606674.
J. W. T. Hessels, S. M. Ransom, I. H. Stairs, P. C. C. Freire, V. M. Kaspi, and F. Camilo. A Radio
Pulsar Spinning at 716 Hz. Science 311 1901–1904, 2006. astro-ph/0601337.
A. Hewish, S. J. Bell, J. D. Pilkington, P. F. Scott, and R. A. Collins. Observation of a Rapidly
Pulsating Radio Source. Nature 217 709–+, 1968.
S. D. H. Hsu. Neutron star vortex dynamics and magnetic field decay: Implications for high density
nuclear matter. Physics Letters B 469 161–165, 1999. nucl-th/9903039.
91
BIBLIOGRAPHY
K. Iida and G. Baym. Superfluid phases of quark matter. III. Supercurrents and vortices. Phys. Rev. D
66 1 014015–+, 2002. hep-ph/0204124.
G. L. Israel, T. Belloni, L. Stella, Y. Rephaeli, D. E. Gruber, P. Casella, S. Dall’Osso, N. Rea, M. Persic, and R. E. Rothschild. The Discovery of Rapid X-Ray Oscillations in the Tail of the SGR
1806-20 Hyperflare. ApJ 628 L53–L56, 2005. astro-ph/0505255.
Y. B. Ivanov, A. S. Khvorostukhin, E. E. Kolomeitsev, V. V. Skokov, V. D. Toneev, and D. N. Voskresensky. Lattice QCD constraints on hybrid and quark stars. Phys. Rev. C 72 2 025804–+, 2005.
astro-ph/0501254.
P. Jaikumar, S. Reddy, and A. W. Steiner. Quark Matter in Neutron Stars:. AN APERÇU. Modern
Physics Letters A 21 1965–1979, 2006a. astro-ph/0608345.
P. Jaikumar, S. Reddy, and A. W. Steiner. Strange Star Surface: A Crust with Nuggets. Physical
Review Letters 96 4 041101–+, 2006b. nucl-th/0507055.
P. B. Jones. Type I and two-gap superconductivity in neutron star magnetism. MNRAS 371 1327–
1333, 2006. astro-ph/0606754.
J. I. Kapusta. Finite Temperature Field Theory. Cambridge University Press, 1989.
C. Kouveliotou, S. Dieters, T. Strohmayer, J. van Paradijs, G. J. Fishman, C. A. Meegan, K. Hurley,
J. Kommers, I. Smith, D. Frail, and T. Murakami. An X-ray pulsar with a superstrong magnetic
field in the soft γ-ray repeater SGR1806 - 20. Nature 393 235–237, 1998.
A. I. Larkin and Y. N. Ovchinnikov. Nonuniform state of superconductors. Zh. Eksp. Teor. Fiz. 47
1136, 1964. Translation: Sov. Phys. JETP 20, 726(1965).
J. M. Lattimer, C. J. Pethick, M. Prakash, and P. Haensel. Direct URCA process in neutron stars.
Phys. Rev. Lett. 66 21 2701–2704, 1991.
Y. Levin. QPOs during magnetar flares are not driven by mechanical normal modes of the crust.
MNRAS 368 L35–L38, 2006. astro-ph/0601020.
B. Link. Constraining Hadronic Superfluidity with Neutron Star Precession. Phys. Rev. Lett. 91 10
101101–+, 2003. astro-ph/0302441.
B. Link. Incompatibility of long-period neutron star precession with creeping neutron vortices. A&A
458 881–884, 2006. astro-ph/0608319.
T. Lippert. 2005. URL http://www.fz-juelich.de/nic/Publikationen/Broschuere/
elementarteilchenphysik-e.html.
A. G. Lyne and F. Graham-Smith. Pulsar Astronomy. Pulsar Astronomy, by Andrew G. Lyne and
Francis Graham-Smith, pp. . ISBN 0521839548. Cambridge, UK: Cambridge University Press,
2005., 2005.
R. N. Manchester. 2006. URL http://www.atnf.csiro.au/news/newsletter/jun06/RRATs.
htm.
R. D. R. Mandal, M. Sinha, M. Bagchi, S. Konar, M. Dey, and J. Dey. Strange pulsar hypothesis.
MNRAS 365 1383–1386, 2006. astro-ph/0511106.
92
BIBLIOGRAPHY
Y. Nambu and G. Jona-Lasinio. Dynamical Model of Elementary Particles Based on an Analogy with
Superconductivity. I. Physical Review 122 345–358, 1961.
T. I. T. (NASA). 2003. URL http://imagine.gsfc.nasa.gov/docs/teachers/elements/
imagine/05.html.
NASA/CXC/SAO. 1999. URL http://chandra.harvard.edu/photo/0052/index.html.
D. J. Nice, E. M. Splaver, I. H. Stairs, O. Löhmer, A. Jessner, M. Kramer, and J. M. Cordes. A 2.1
M Pulsar Measured by Relativistic Orbital Decay. ApJ 634 1242–1249, 2005. astro-ph/0508050.
O. E. Nicotra, M. Baldo, G. F. Burgio, and H. Schulze. Hybrid protoneutron stars with the MIT bag
model. ArXiv Astrophysics e-prints 2006. astro-ph/0608021.
B. Niebergal, R. Ouyed, and D. Leahy. Magnetic Field Decay and Period Evolution of Anomalous
X-Ray Pulsars in the Context of Quark Stars. ApJ 646 L17–L20, 2006. astro-ph/0603741.
J. R. Oppenheimer and G. M. Volkoff. On Massive Neutron Cores. Phys. Rev. 55 4 374–381, 1939.
D. Page, U. Geppert, and F. Weber. The cooling of compact stars. Nuclear Physics A 777 497–530,
2006. astro-ph/0508056.
Particle Data Group. Standard model and related topics. Physics Letters B 592 104–185, 2004.
M. E. Peskin and D. V. Schroeder. An Introduction to Quantum Field Theory. Westview Press, 1995.
J. D. H. Pilkington, A. Hewish, S. J. Bell, and T. W. Cole. Observations of some further Pulsed Radio
Sources. Nature 218 126–129, 1968.
W. H. Press, S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery. Numerical recipes in C. The art of
scientific computing. Cambridge: University Press, second, repr. edition, 1996.
C. Price. Neutrino bremsstrahlung from quarks and the cooling of neutron stars. Phys. Rev. D 22
1910–1914, 1980.
K. Rajagopal and R. Sharma. Crystallography of three-flavor quark matter. Phys. Rev. D 74 9 094019–
+, 2006. hep-ph/0605316.
K. Rajagopal and F. Wilczek. The Condensed Matter Physics of QCD. ArXiv High Energy Physics Phenomenology e-prints 2000. hep-ph/0011333.
S. Reddy, M. Prakash, and J. M. Lattimer. Neutrino interactions in hot and dense matter. Phys. Rev. D
58 1 013009–+, 1998. astro-ph/9710115.
S. B. Rüster and D. H. Rischke. Effect of color superconductivity on the mass and radius of a quark
star. Phys. Rev. D 69 4 045011–+, 2004. nucl-th/0309022.
E. Salpeter. The luminosity function and stellar evolution. ApJ 121 161–+, 1955.
A. Sedrakian and J. W. Clark. Nuclear Superconductivity in Compact Stars: BCS Theory and Beyond.
ArXiv Nuclear Theory e-prints 2006. nucl-th/0607028.
T. V. Shabanova, A. G. Lyne, and J. O. Urama. Evidence for Free Precession in the Pulsar B1642-03.
ApJ 552 321–325, 2001. astro-ph/0101282.
93
BIBLIOGRAPHY
J. Shaham. Free precession of neutron stars - Role of possible vortex pinning. ApJ 214 251–260,
1977.
S. L. Shapiro and S. A. Teukolsky. Black Holes, White Dwarfs and Neutron Stars: The Physics of
Compact Objects. Black Holes, White Dwarfs and Neutron Stars: The Physics of Compact Objects,
by Stuart L. Shapiro, Saul A. Teukolsky, pp. 672. ISBN 0-471-87316-0. Wiley-VCH , June 1986.,
1986.
I. H. Stairs, A. G. Lyne, and S. L. Shemar. Evidence for free precession in a pulsar. Nature 406
484–486, 2000.
Star Collaboration. Experimental and theoretical challenges in the search for the quark gluon plasma:
The STAR Collaboration’s critical assessment of the evidence from RHIC collisions. Nuclear
Physics A 757 102–183, 2005. nucl-ex/0501009.
M. Stejner and J. Madsen. Gaps below strange star crusts. ArXiv Astrophysics e-prints 2005. astroph/0512144.
T. E. Strohmayer and A. L. Watts. Discovery of Fast X-Ray Oscillations during the 1998 Giant Flare
from SGR 1900+14. ApJ 632 L111–L114, 2005. astro-ph/0508206.
S. E. Thorsett and D. Chakrabarty. Neutron Star Mass Measurements. I. Radio Pulsars. ApJ 512
288–299, 1999. astro-ph/9803260.
M. Tinkham. Introduction to Superconductivity. McGraw-Hill, second edition, 1996.
R. C. Tolman. Static Solutions of Einstein’s Field Equations for Spheres of Fluid. Phys. Rev. 55 4
364–373, 1939.
N. Tominaga, M. Tanaka, K. Nomoto, P. A. Mazzali, J. Deng, K. Maeda, H. Umeda, M. Modjaz,
M. Hicken, P. Challis, R. P. Kirshner, W. M. Wood-Vasey, C. H. Blake, J. S. Bloom, M. F. Skrutskie, A. Szentgyorgyi, E. E. Falco, N. Inada, T. Minezaki, Y. Yoshii, K. Kawabata, M. Iye, G. C.
Anupama, D. K. Sahu, and T. P. Prabhu. The Unique Type Ib Supernova 2005bf: A WN Star Explosion Model for Peculiar Light Curves and Spectra. ApJ 633 L97–L100, 2005. astro-ph/0509557.
UNO collaboration. Physics Potential and Feasibility of UNO. 2001. URL http://ale.physics.
sunysb.edu/nngroup/uno/whitepaper/index.html.
V. V. Usov. Thermal Emission from Bare Quark Matter Surfaces of Hot Strange Stars. ApJ 550
L179–L182, 2001. astro-ph/0103361.
H. J. Warringa. The phase diagram of neutral quark matter with pseudoscalar condensates in the colorflavor locked phase. ArXiv High Energy Physics - Phenomenology e-prints 2006. hep-ph/0606063.
A. L. Watts and T. E. Strohmayer. Detection with RHESSI of High-Frequency X-Ray Oscillations in
the Tailof the 2004 Hyperflare from SGR 1806-20. ApJ 637 L117–L120, 2006. astro-ph/0512630.
F. Weber, M. Meixner, R. P. Negreiros, and M. Malheiro. Ultra-Dense Neutron Star Matter, Strange
Quark Stars, and the Nuclear Equation of State. ArXiv Astrophysics e-prints 2006. astroph/0606093.
E. Witten. Cosmic separation of phases. Phys. Rev. D 30 272–285, 1984.
94
BIBLIOGRAPHY
A. Wolszczan and D. A. Frail. A planetary system around the millisecond pulsar PSR1257 + 12.
Nature 355 145–147, 1992.
95